Pre-modern era identifiers in Japanese calendar

[ptomato]

Related to Temporal, discussed in tc39/proposal-temporal#526. Eras in a calendar must have a unique identifier.

From tc39/proposal-temporal#526 (comment):

We had a lengthy discussion about this today and decided to use the transliteration as the era identifier for the five modern eras (starting with meiji). For pre-meiji eras, we will use either a century syntax (like heisei-13xx) or just fall back to the Gregorian CE/BCE eras.

Some user research is probably indicated here, to determine whether there is an existing convention for this.

[justingrant]

FYI, "just fall back to the Gregorian CE/BCE eras" is what's currently implemented in the Temporal polyfill. If the 402 WG decides on the other solution, then we'll need a Temporal PR to make the change.

[justingrant]

@Manishearth @sffc - what's the plan to resolve this issue?

[justingrant]

@Manishearth @sffc - what's the plan to resolve this issue?

Oh, and while we're at it, here's the full list of era issues that I know of. When do these need to be resolved to ensure that we're not stuck supporting whatever the initial browser implementation does for web-compatibility reasons?

• Pre-modern era identifiers in Japanese calendar #541 (this issue)
• Which era should be the year anchor for the ethiopic calendar? #534
• ICU calendars - how to handle {constrain, reject} for out-of-range {eraYear, era} value pairs? #540
• Hebrew calendar - support or use a constant era? #535

[Manishearth]

I think the final proposal we settled on was "Year, post-meiji elision+truncation (reiwa, showa-13xx)". This means that:

• Eras are represented as era-year
• year is eliminated for Meiji onwards, (meiji, showa, heisei, reiwa)
• The last two digits of year are truncated (showa-13xx, etc) if it hasn't been eliminated

So basically pre-modern eras are represented as era-yyxx, with yy being the thousands and hundreds digits of the year.

[justingrant]

@Manishearth - Also, what's the plan for years before the first Japanese era? Will those dates get the 'ce' era? Will they throw? Use negative eraYear values and the oldest Japanese era?

• year is eliminated for Meiji onwards, (meiji, showa, heisei, reiwa)

Don't forget Taishō! (1912–1926 CE)

• The last two digits of year are truncated (showa-13xx, etc) if it hasn't been eliminated

What does "if it hasn't been eliminated" mean?

Also, what happens if there's a historical revision in the dates assigned to an era, e.g. if new primary sources are discovered that move the year (or, rarely, the century) of the start of an era?

[Manishearth]

Don't forget Taishō! (1912–1926 CE)

Right, it was a list of examples.

What does "if it hasn't been eliminated" mean?

"If year hasn't been eliminated entirely because the era is a modern one". Basically, pre-modern eras have years with the last two digits replaced with xx, modern eras do not have years

(unless the modern era is a duplicate name of another modern era)

Also, what happens if there's a historical revision in the dates assigned to an era, e.g. if new primary sources are discovered that move the year (or, rarely, the century) of the start of an era?

The xx thing handles the year moving around. Ultimately the mapping is just a mapping, so if the century changes, the mapping will be less logical, but it will not be less useful, and it's rare enough that that shouldn't be a huge deal overall.

[ptomato]

Making a note here that part of the resolution of this issue needs to be to revert the removal of tests in tc39/test262#3320