Repository navigation
Validation and Release Status
Evidence checkpoint: 2 November 2026, public content deployed at 03:24 and private Preview restored at 03:28 Europe/Rome. PR #37 is integrated into main at d2525bf. Future runs and remaining qualification checks are not counted in advance.
| Surface or check | Observed result | What it does not establish |
|---|---|---|
| Public service | Stable product 5.5.0; see the 2 November release receipt | Engine 6 is not activated there |
| Public notebook | Current homepage access banner and bilingual diary published on 2 November; six diagrams | A notebook is not an interactive preview |
| Source integration | PR #37 merged to main | Online runtime availability |
| PR #39 parser correction, not hosted | 1,569 local tests passed, 14 skipped; six GitHub checks passed on that candidate | Source completeness or a live campaign day |
| Candidate 3 automated suite | 1,565 passed, 14 skipped | Live retrieval quality or complete coverage |
| Candidate 3 generated standalone | 22 checks passed on the actual generated server; fresh isolated database | Successful managed-host deployment |
| Candidate 2 HTTP/worker rehearsal | 44 checks passed; real Apple capture, private comparison and process-crash recovery | Candidate 3 rerun of those 44 checks, or interrupted-network-request recovery |
| Earlier candidate 2 stripped-source rehearsal | 18 checks passed without src/scripts | The later generated-standalone test is a separate execution |
| Original channel tests | 14 passed; 12 concurrent-start iterations | Universal concurrency or VPS capacity |
| Original browser flow | Selector, private report, differences and JSON download verified with two synthetic policies | Real-source completeness |
| Hosted deployment | Second upload Completed / Current at 23:17 Europe/Rome; private console online | Long-lived managed-process availability |
| Hosted runtime verification | 47 checks passed; real Apple comparison completed, three missing-contract holds, official-state digests unchanged | Complete source coverage or 500-source capacity |
| Hosted browser verification | Administrator login, hydrated selector and saved report verified; 22,649-byte JSON export verified on 2 November against the saved API report | The browser event observer timed out despite a completed verified file |
| Current parser daily cohort | Frozen PR #39 candidate; 12 sources, 1 accepted live date (2 November) | At least 3 dates and 48 actual hours are still required |
| Previous parser daily cohort | Preserved: 12 sources, 2 accepted dates, minimum span 20.7712 hours | Incomplete; cannot qualify the changed parser |
| Previous parser daily text comparison | All 12 extracted hashes unchanged; 10 payloads differ | Equivalence of legal meaning or complete extraction |
| Real catalogue probe | 113 configured policies, 104 distinct HTTP addresses, 107 attempts | 500 distinct live sources |
| Synthetic capacity | 500 records tested with persistence/recovery work | Real Internet or production-infrastructure throughput |
| Source completeness | One bounded Codex-assisted hosted smoke contract; other source, structural and PDF-language cases remain held | Human-certified or universal completeness |
| Earlier operational public check | 9 passed, 1 external; health and integrity ok | Independent dynamic security audit remains external |
The frozen candidate attempted a second daily window at 21:35 Europe/Rome. 11/12 sources returned HTTP 200; Meta timed out on both direct transport and HTTP/2 fallback. All 13 attempts are retained. The final campaign assessment rejected the window, so the count remains 1 accepted day of 3. A zero process exit and 12 processed source rows are not proof of 12 successful captures.
Ten comparable extracted-text hashes match the first window. AWS differs at its update date and section 83; the bounded snapshot comparison remains held, without completeness approval or a published policy-change notice. A third valid date cannot occur before 5 November, subject to actual 20-hour and 48-hour checks. Execution and failure receipt.
The content-only website update is prepared and locally built on the stable 5.5 line. Its publication is pending staging and production verification; the online diary must not yet be described as updated.
At 21:03 Europe/Rome, the frozen table-escaping candidate 050af19 completed its first accepted live window: 12/12 HTTP 200, 12 attempts, no retries, 35.287 seconds. The manifest and implementation stayed unchanged. This is one counted date and zero within-cohort span hours, distinct from the earlier parser, main and the hosted Preview. AWS PDF language remains unverified, and HTTP success does not establish source completeness or origin freshness. Cache metadata is retained in the dated execution receipt.
The next all-source 20-hour threshold is 3 November at 17:03:41 Europe/Rome. The third window must start after 4 November 21:03:43 Europe/Rome to satisfy 48 actual hours; a 21:00 start would be premature. The daily automation now retains that timestamp guard. Independent completeness annotation and realistic 500-source capacity remain open; no new Gemini/GLM execution is claimed.
The new day-1 evidence is now visible in the bilingual public diary. The same immutable archive passed staging before production: 18 EN/IT page checks and 30 source hashes matched in each environment. Public product 5.5.0 retains the established scanner, 35 application tables and 19 migrations. Nine internal runtime checks passed; the independent security check remains external, so overall readiness is still attention.
The exact private Preview r2 was restored at 22:25 Europe/Rome, with all 12 restoration checks passing and all 12 monitored table hashes unchanged. Its saved report remained completed and private. No hosted scan or extra daily date was added. Failed HTTP/upload attempts and a Vivaldi screenshot anomaly are retained; the public diary was visually confirmed in an independent browser. Immutable publication and restoration receipt.
Published at 04:33 Europe/Rome on 2 November after staging verification. The exact same archive passed 49 production release checks, all 595 sitemap destinations and 30 source-file hash comparisons. The previous private Preview was restored at 04:38, passed 12 restoration checks, and preserved the full-row hashes of all 12 observed tables. No new live capture or daily window was counted.
The 5.5 release announcement links the private engine workspace, source-contract review, extraction comparison and JSON reports. Homepage, changelog, roadmap, both atlases, newsroom feeds and current EN/IT fact sheets identify product 5.5.0. Engine 6.0 is its experimental engine track, hosted separately with administrator authentication and private results.
The public artifact uses the established scanner. The Engine 6 source distribution on main stays restricted to private staging independently of the product number; neither a release label nor a selector enables public policy changes or notifications. Historical package versions and dated receipts below remain unchanged. Release validation and deployment receipt.
At the earlier 03:24 Europe/Rome checkpoint on 2 November, the public homepage access banner and bilingual notebook were published through staging and an identical production archive, retaining runtime 5.2.0. Staging passed 11 required checks plus 22 content checks; production passed 22 content checks and nine internal runtime checks. The independent security test remains external. The same private Preview was restored at 03:28 Europe/Rome: 12 restoration checks passed and full-row digests across 12 observed tables stayed unchanged. The existing report was exported in the browser: its 22,649-byte JSON file equals the saved API response. No new capture or daily window was counted. Publication and restoration receipt.
The actual Hostinger run 4e9eef90-0bbb-42ee-b945-922ac76ffffc completed across four selected policies. Apple was acquired directly with HTTP 200 at 23:20:48 Europe/Rome, yielding 28,124 Preview characters and 2,089 stable-extractor characters on the same response. Outcome: confirmation_pending; no AI analysis or publication. Three sources were held for missing contracts. Official policy, history, check-log, source-contract, baseline, subscriber and outbox row digests remained unchanged.
The hosted HTTP process and worker reported implementation hash d9d5dde4f1c8d22c03c80ae200bc3e44734ef6a430c7e7bd037719ee30614e88. Archive SHA-256: c887cd49959f38ab4137c235349c1f2e5785fed97848a4a42daf4410ef24bf26; source: 34c188c0d54ec24d61469ca1683dec8988252c6f; Hostinger deployment: 01a1992e-99e7-712f-bbfc-3bf04cb30890.
One earlier initial GET timed out after 30 seconds and remains retained. A follow-up GET returned the expected 307 in 576 ms. The subsequent complete hosted suite passed 47/47; it was not repeated or counted twice. Hosted restart recovery was not tested by restarting the service. Browser login, hydrated channel selection and the completed saved report have separate evidence. The report rendered one different result, three unavailable comparisons and 2,089 versus 28,124 Apple characters. The original browser JSON attempt remained unverified at that checkpoint. The 2 November export verification above supersedes that open item while retaining the observer timeouts. The authenticated API report and downloaded file were compared separately.
Candidate 2 retrieved the official Apple HTML response and applied two extractors to that same capture: 28,124 versus 2,089 characters. The outcome was confirmation_pending, with no AI analysis or public policy update. This is extractor divergence, not a detected provider change. Process-crash recovery occurred after that capture completed and waited for actual leadership expiry. Full-row invariants preserved official baseline/history/log and notification/subscriber state.
Candidate 3 corrected a real Hostinger build failure: the provider's next.config.ts wrapper changed the implementation hash. The approved normalizer accepts only the exact observed wrapper and manifest-matching base, then restores approved bytes before preflight. The first failed deployment and failed harness attempts remain recorded. The successful 22-check generated-standalone run preserved the actual Next payload, including nine legitimately traced adaptive-acquisition auxiliary files, and confirmed that engine implementation sources were absent.
The hosted candidate's 1,565-test full suite supersedes earlier local checkpoints of 1,469 and 1,536 passing tests; those historical receipts are retained. Test totals are not added together to imply independent coverage. Three local-tool filesystem races were repaired and tested; seven medium CodeQL data-flow findings have separate triage. This checkpoint does not assert a fresh passing remote security gate.
The corpus and acquisition implementation identity are frozen. Count only completed live captures on distinct Europe/Rome dates, with at least 20 elapsed hours per source between accepted windows and at least 48 hours from first to last. Retries, replay and HTTP 304 do not supply additional days. The next invocation may legitimately leave the count unchanged.
The previous parser's first two windows are 31 October at about 23:42 and 1 November at about 20:28 Europe/Rome. The PR #39 table-encoding correction changed the extraction identity, so its new frozen 12-source cohort starts from zero live windows. The first observation completed on 2 November at 21:03 Europe/Rome, as recorded above. The PDF language check for AWS remains unverified. Two days of unchanged text are a bounded observation, not a completeness certificate. Hosted smoke captures belong to a separate corpus and do not alter or advance this frozen daily cohort.
Gemini and GLM supplied reviews recorded in the reports. Their opinions are not execution receipts. The hosted Gemini session displayed Argon Alta; it must not be presented as a verified Gemini 4 model identity. GLM-5.3 Max reviewed frozen candidates in ZCode Plan mode without executing tests. Its candidate 3 review found no concrete integrity bypass or runtime mismatch and retained the managed-host deployment requirement.
Locally verified corrections include preservation of Policy timestamps, isolated report reads, worker preflight, immutable privacy identity, lease renewal and terminal-run recovery. An interrupted or quota-switched model response is documented as such.
The revision 03 diary ZIP SHA-256 is 1ca49fa81461399f79c349b9404fd816f261050249d62ff25562453c48c8129c, source revision 5e2cb71324e795d418eef1b6122cfc384fb5b899. Staging passed 11 operational and 10 page checks; production passed 11 page checks. That deployed package changed the diary and release manifest only, retaining runtime 5.2.0. Raw captures, private configuration and databases are excluded from Git; committed manifests record artifact hashes.
Documentation checkpoint: 2 November 2026. Product 5.5.0 online; Engine 6 source on main; private Preview online with administrator authentication. Stable service · Engine 6 notebook · Validation and Release Status