Methodology

The IANA Time Zone Database, read through the JavaScript Intl API at build time. No offset is written by hand anywhere in this site. That matters because zone rules change — Mexico abolished daylight saving in 2022, Egypt reintroduced it in 2023, Jordan moved to a permanent UTC+3 in 2022 — and a site holding its own copy of those rules goes quietly wrong.

A wall-clock time is first resolved to a real instant in the source zone, then that instant is formatted in the target zone. Resolving the instant takes two passes: guess using the offset at the naive UTC interpretation, then re-check against the offset actually in force at that guess. Converting by adding a fixed offset instead is what produces answers that are an hour wrong for half the year.

Two awkward cases, both handled explicitly. A wall time inside the spring-forward gap never happens — 02:30 on the US switch day does not exist — and resolves forward to 03:30. A wall time inside the autumn fall-back overlap happens twice, and resolves to the first, earlier occurrence. These are choices, not accidents, and both are covered by tests.

Because they genuinely have no widely-used abbreviation, or because the obvious one collides with a better-known zone. Bangladesh Standard Time is officially BST, which is also British Summer Time; Taiwan uses CST, which is also China and US Central. In those cases the numeric form is shown. An honest "+06" is better than a familiar abbreviation that means something else.

A 9:00–18:00 day in India is compared against 9:00–17:00 elsewhere, reflecting the actual norm difference. An hour counts as shared only if the whole hour falls inside both working days — with India’s half-hour offset, rounding here is exactly how wrong answers get published. Each side’s weekday is read from its own clock, because a meeting slot can sit on different calendar days at each end.

Three sources, each labelled on the page. Published calendars via https://date.nager.at for 55 countries, last refreshed 2026-09-12. Hand-curated national holidays for the 13 countries that source does not cover — India among them. And holidays computed from their own calendars: Eid from the Umm al-Qura calendar, Chinese New Year from the Chinese calendar, the Jewish holidays from the Hebrew calendar, and Good Friday from the Easter computus. Anything lunar is marked predicted, because the observed date can move once announced.

Indian state holidays, Sri Lankan Poya days, Nepali Bikram Sambat festivals, Malaysian state holidays, and employer-specific calendars. These need a maintained regional dataset. Pages say what they exclude rather than implying completeness, and conversions between two places that each span many countries show no holiday panel at all rather than picking one country’s calendar.

Roughly twice a year, matching the cadence of IANA database releases, plus whenever content changes. Because pages are generated from the runtime time zone database at build time, a rebuild is what carries a rule change onto the pages.