Maintenance
Part of Country calendars: three origins of a closed day
Country calendars holidays: a review routine for each new national closure list
A step-by-step routine for reviewing a new national closure list against the old one, so country calendars holidays changes never move a count unseen.
A new edition of a national closure list is a change request, not a refresh. Accept it unreviewed and an old count can move with nothing to say why. A short routine, run once per edition, keeps every answer traceable to a cause.
No country's closures or dates are listed here. A list with no publisher and no issue date cannot be used in a dispute, so the routine works on whatever list you hold, for any country, and carries over from one year to the next.
What to take away
- Compare each new edition with the one it replaces, entry by entry, before any saved count is rerun.
- Sort every difference by kindfixed date, rule based, or announced. The kind decides the action.
- Record the list version apart from the arithmetic, so a changed count has one named cause.
- Avoid the country calendar mistakes that start with a copied date: recompute rules, do not recycle their results.
The review in six steps
Run the same six steps for every edition, in this order. Each step narrows what the next one has to check, and the last one proves the rest.
- Fetch the new edition from the body that publishes it and note the day you retrieved it.
- Find the enacting instrument behind it and read that, not a summary.
- Compare the two editions entry by entry and sort each difference.
- Check the scopewhether the list binds courts, banks, public offices, or all three.
- Check the substitution rulewhether a closure on a rest day moves to a working day.
- Rerun a few saved counts under both versions and file the difference.
Step three produces the change list and step six tests it. A count that moves only on a date the comparison already explained is a clean upgrade. A count that moves anywhere else points to a missed difference or to the arithmetic.
Sorting what changed
Every difference falls into one of three kinds of entry, and the kind decides what you do with it. Fixed dates are read off the instrument. Rule-based entries are recomputed. Announced entries are never assumed from a list published long before the day.
Three kinds of calendar entry
Fixed date
- Fixed by
- Statute names date
- Known a year ahead
- Yes
- Planning risk
- Low
Rule based
- Fixed by
- Statute names rule
- Known a year ahead
- Yes, by computing
- Planning risk
- Breaks if copied
Announced
- Fixed by
- Discretionary proclamation
- Known a year ahead
- No
- Planning risk
- Breaks silently
| What you see in the new edition | Possible cause | Action |
|---|---|---|
| A fixed-date entry shows another date | The instrument was amended, or the old list was wrong | Read the instrument, then take its date |
| A rule-based entry resolves to a different day | The rule changed, or the calendar and the rule interact differently this year | Recompute from the rule |
| An entry appears with no instrument behind it | An announced closure | Add it with its announcement as the source |
| An entry disappears | It was dropped, or moved to another scope | Confirm in the instrument before removing it |
| Only a name differs | Publishers name days differently | Keep the stored date and note the name used |
Rule-based entries need the most care. Copying last year's resolved dates forward turns a rule-based entry into a fixed one and quietly breaks it. Implement the rule from the instrument, not the date it produced last time.
Step two in practice: the instrument
The instrument decides the edge cases, and a summary drops the parts that decide them. In the United States the federal set is codified at 5 U.S.C. 6103, and the enacted text is published through official channels, including the Government Publishing Office's Statutes at Large collection.
Other countries use a statute, a ministerial order, a proclamation power, or a combination. Read whichever one applies.
Names need the same care. OPM says it always refers to holidays by the names designated in the law, although state and local governments and private businesses may use other names. A list that stores only names leaves the resolving to the reader.
The instrument alone answers the scope and substitution questions. Who may add a closure during the year? Does it bind everyone, or only one class of body? A day on which government offices close is not automatically a day on which a payment cannot settle, and a list that merges the two is wrong for both purposes.
Recording the outcome
File the review as a release note kept beside the list, not in someone's inbox. Hold these fields with every edition, because each one prevents a specific argument later.
Fields to hold beside a list
- Publisherwhich list governs
- Instrumentsource, not summary
- Scopecourts vs settlement
- Effective periodno silent reuse
- Retrieved onstale vs broken
- Substitution ruleweekend handling
- Versionsame list, same meaning
- Publisher, so two lists that disagree can be told apart
- Instrument, so a summary is not treated as the source
- Scope, so a courts list is not applied to settlement, or the reverse
- Effective period, so last year's list is not reused without a decision
- Retrieved on, so a stale answer is not read as a broken formula
- Substitution rule, so weekend closures are not deducted by accident
- Version, so two people quoting one list mean the same thing
Treat the update like a software release: consumers pin to a version, an upgrade is deliberate, and the previous version stays answerable. A list edited in place is a release with no number. Old counts can then be reproduced from the version that produced them.
Feeding the result into counts
Keep the weekday base and the closure deduction as separate numbers. The base changes because the year starts on a different day; the deduction changes because the list changed. A single total hides which happened, and the two have different owners.
The base is worked out under counting a year. Store the resolved dates with their version, never the names. A name has to be resolved by a rule before it can be subtracted, and that rule is the part that differs between calendars.
A stored date carries a resolution already made by someone entitled to make it. The sub-national layer that sits on top of a national list is under state and regional calendars.
Where the rest-day convention itself differs, the arithmetic is under how national calendars differ. The day-stepping form, useful for step six, is under adding business days.
Common questions
Why does the routine list no country's 2027 closures?
Because the list that governs your deadline depends on which body set it. One printed here would have no publisher and no issue date, and would be wrong for most readers with nothing to say which.
Can I compute a rule-based closure myself?
Yes, if you implement the rule from the instrument rather than copying a resolved date. That is the only version that stays correct in the year the rule and the calendar interact differently.
How should an announced closure be handled in a plan made a year ahead?
As a known unknown. Record that the list may gain entries, and recompute rather than assume the plan's figure is final. When the announcement arrives, run the routine on that one entry.
Is a widely used published list good enough?
For most operational purposes, yes, provided you record which one you used and when you took it. It stops being good enough the moment the number is disputed.
