Stage 2.7 review feedback: Be more specific about calendar authorities

[sffc]

There are multiple potential reckonings for the Chinese calendar, the Islamic calendar, the Thai Solar (Buddhist) calendar, the Japanese calendar, and others.

For these calendars, the spec currently states:

Calendar	Description
"buddhist"	Thai Buddhist calendar, proleptic. Month numbers, month codes, and days are the same as in the ISO 8601 calendar, but the epoch year is different. There is one era.
"chinese"	Traditional Chinese calendar, proleptic. Similar lunisolar algorithm to "dangi". The arithmetic year is identical to "gregory" and there are no eras.
"islamic-umalqura"	Hijri calendar, proleptic, Umm al-Qura. Lunisolar calendar using KACST-calculated months from the start of 1300 AH to the end of 1600 AH and falls back to "islamic-civil" outside that range.
"japanese"	Japanese Imperial calendar, era system hybridised with "gregory". Month numbers, month codes, and days are the same as in the ISO 8601 calendar, extended proleptically before their introduction in ISO year 1873. Imperial era names only extend as far back as the Meiji period (starting in ISO year 1868) during which calendar reforms took place. The arithmetic year, and the years and eras before ISO year 1868, are identical to "gregory".

The description of "islamic-umalqura" is excellent: it says exactly what authority to use for in-range dates, and what to do for the proleptic approximation dates.

Likewise, "japanese" is quite good. It would be slightly better if it stated an authority of where to source the era names and cutoff dates, but at least it tells you that it aligns with ISO-8601 and is used proleptically before 1868.

The row on "buddhist" looks okay on the surface, until you learn about how the current proleptic rules were not adopted until 1940¹. If we wish to support the calendar extending before 1940, then we need to be more clear about this in the description.

The row on "chinese" ought to disambiguate the astronomical authority. The Chinese government has a standard GB/T 33661-2017, which likely is the one that we should reference², and that implementations like ICU4X should reference. This solves the problem for modern dates, but it was not long ago that the Chinese calendar was using other methods like non-UTC time and the píngqì approximation³.

The "dangi" calendar also probably needs more specificity, but I haven't done the research to determine what authority we should be using there.

Footnotes

1. https://unicode-org.atlassian.net/browse/ICU-23162 ↩

2. https://ytliu0.github.io/ChineseCalendar/rules.html ↩

3. https://ytliu0.github.io/ChineseCalendar/solarTerms.html ↩

[sffc]

It appears that the authority for the "dangi" calendar is the Korea Astronomy and Space Science Institute (KASI).

https://astro.kasi.re.kr/

As mentioned in #64, we should cite the authority, but we should also state the ranges where we refer to the authority. This particular authority publishes data through 2050.

[ptomato]

Do we leave this open for buddhist? I think the PR resolved everything else, except possibly the external authority for the Meiji and following eras.

[sffc]

I haven't researched the answer yet for Buddhist.

[sffc]

Just dumping some more useful links in here:

The GB/T standard (here) refers to IERS for astronomical calculations.

The Hong Kong Observatory, though not the official authority, publishes machine-readable tables that appear to align with the GB/T standard. The Purple Mountain Observatory is the official authority, and they publish a PDF. Also see unicode-org/icu4x#6454.

See unicode-org/icu4x#6455 for more on the Dangi calculations.

[ben-allen]

I've put up a draft PR adding text to the description of the Buddhist calendar indicating the complexities that arise for dates before 1941.