Complete the East-Asian calendar matrix

[robertbastian]

	Japan	Korea	China
Gregorian months, regnal year	-u-ca-japanese		-u-ca-roc1
Lunar months, cyclic year		-u-ca-dangi	-u-ca-chinese
Lunar months, regnal year			2

We might want to add Japanese calendars with lunar months (cyclic and regnal), as well as a Chinese lunar calendar that uses the ROC era.

We should consider this as part of #6938. In particular we need a way to link our existing Japanese calendar (Gregorian months, regnal years) to a potential Japanese lunar calendar.

Footnotes

1. Kind of, the ROC era is in the tradition of the Chinese imperial eras. We might want to support both imperial eras and BROC. ↩

2. This would be the imperial eras followed by the ROC era ↩

[Manishearth]

This would impact the trait in #6938 and make it need era info. This might be complicated, not just a matter of a single to/from era API, depending on @sffc's from_fields work.

My inclination is to not complicate things this way unless we have user need. We could introduce a ChineseWithRegnal trait that inherits from the one in #6938 and have a separate type. Not great, not terrible.

[hsivonen]

We shouldn't do stuff "for completeness". That's how things like h24 happen.

Let's not add calendars that haven't had the kind of civil use demand to make it into CLDR.

[robertbastian]

If I understand your comment in #6492 (comment) correctly then .NET has a lunar calendar with ROC years. They probably had some reason to add it.

Let's not add calendars that haven't had the kind of civil use demand to make it into CLDR.

We already have two of those, Julian and simulated Hijri.

[Manishearth]

Yeah, but we don't know if it's a good reason. Worth seeing who the users are.

[robertbastian]

I mean it's similar for the Hijri calendars, I feel like we only implement all the variants because some system before us did.

[Manishearth]

Sure, but it's not "some system", it's ICU4C and the LDML/Unicode standard; we are a Unicode project and to some extent aiming for compatibility.

[robertbastian]

By "we" I mean ICU/LDML. To this day I have not found anyone who can tell me where all of these are used (and what rgsa is even supposed to be).

[sffc]

We already have two of those, Julian and simulated Hijri.

Julian is used in ICU but not ICU4X for doing Old System.

Simulated Hijri is useful in applications that allow the user to tweak their observations. Microsoft had this feature at some point (or maybe still does).

RGSA is "ReGion Saudi Arabia" for the Hijri sighting. If someone wanted to add an Egypt sighting, it would be RGEG.

I'm ICU4X we don't ship things that have no use case. This is a departure from what CLDR has sometimes done but aligns with the Web.

[robertbastian]

Julian is used in ICU but not ICU4X for doing Old System.

So it's not used by ICU4X or ICU4X clients.

RGSA is "ReGion Saudi Arabia" for the Hijri sighting.

But what is that meant to mean? SA doesn't use a sighting other than UAQ, so this is just a calendar that someone at some point asked for, CLDR added, but nobody knows how to implement and it's just carried along. https://cldr.unicode.org/development/development-process/design-proposals/islamic-calendar-types