Holiday API vs Official Calendars: What Was Actually Verified

The verified bundled combinations are US 2026, US 2027, GB 2026, SG 2026, and SG 2027. Each passed only for the scope stated in the ledger. GB 2027, KR 2026, JP 2026, DE 2026, AU 2026, and CA 2026 were not promoted to verified snapshots.

A result marked not verified or scope incomplete can still contain useful API rows. It cannot be represented as a manually confirmed bundled calendar. The resolver may show live API data with an unverified warning, or an unsupported or unavailable state when reliable data is not available.

On July 28, 2026 in Asia/Seoul, each listed Nager.Date payload was collected and its HTTP status, raw character length, and UTF-8 SHA-256 recorded. National and subdivision rows were separated before dates and scope were compared with the current official primary-source page named in the ledger.

Every request in the raw access register returned HTTP 200. That transport result did not decide verification. The country comparison and scope decision controlled whether a snapshot could be bundled.

The US records match the 11 observed federal closure dates for Monday-Friday federal employees in 2026 and 2027; regional API rows were excluded. GB 2026 matches the official lists and division assignment for England and Wales, Scotland, and Northern Ireland. Singapore 2026 and 2027 match the gazetted dates represented as closure dates under the Ministry of Manpower's stated common Sunday rest-day substitution pattern.

None of those decisions proves a complete private-sector calendar. State, local, bank, school, employer, leave-entitlement, and alternative rest-day rules remain outside the stated verified scope.

GB 2027 contains a direct Scotland substitute-day mismatch. Japan's API payload omits two official 2026 dates and mislabels the May 6 substitute holiday. South Korea's API scope differs from the older official calendar-standard announcement and later changes complicate the comparison. Canada compares an API holiday classification with a CRA deadline calendar whose purpose is different.

Germany and Australia failed on completeness rather than a countrywide verification claim. Berlin and the New South Wales sample were compared, but the remaining jurisdictions were not fully checked. A partial sample cannot support a national snapshot.

HTTP failure, blocked fetch, invalid JSON, an invalid payload, or an unverified empty array must never become a successful zero-holiday calendar. The resolver prefers a verified snapshot, then fresh cache, then live API data, and otherwise returns an explicit unsupported or unavailable state.

India is explicitly unsupported in the current resolver. A prior empty response is not interpreted as proof that a year has no holidays. Users are directed to the relevant official national, state, bank, or organization calendar.

Treat live rows as a candidate list. Select the jurisdiction and operational scope, compare every date with the current official source, separate actual and observed dates, and add bank, school, employer, carrier, or organization closures that the obligation recognizes. Record the comparison date and mismatches.

Do not change an unverified label merely because the list looks plausible or because an API call returned 200. Verification requires a complete comparison for the scope being claimed.

A correction report should include country, year, selected scope, API URL, affected row, official URL, access date, expected date or classification, and the observed mismatch. Send non-sensitive corrections to [email protected].

The raw lengths and hashes in the ledger identify the payloads collected on July 28, 2026. A later payload can be compared without pretending it is the same artifact. Review and source-verification dates change only when the underlying work is repeated.

The ledger is a dated comparison, not a permanent certification of an API, country, or year. Temporary holidays, later legal changes, regional decisions, employer policies, banking schedules, school calendars, and provider operations can change after the comparison.

Only the scopes and combinations marked verified are bundled. Do not infer verification for another year, another subdivision, another rest-day pattern, or a country-level calendar from a sample jurisdiction.