Where should the extended year be anchored in `ethiopic`?

[sffc]

Currently, I believe we anchor it to the Amete Mihret era.

@robertbastian suggested in unicode-org/cldr#4519 (comment) that we could use the Amete Alem as the epoch for both ethiopic and ethiopic-amete-alem.

I think it still makes more sense to set the epoch of ethiopic to Amete Mihret since that is the most widely understood year number. I don't really see the advantage of making it be Amete Alem.

Let's decide.

[Manishearth]

I think it's weird for extendedYear to be something other than a valid eraYear "default" era. Making them both default to Amete Alem seems fine since both calendars have it. Having them use different extended years is also fine, and that preserves Amete Mihret for the ethiopic calendar, which as you say makes more sense.

[sffc]

For sure Amete Mihret will be the displayed era when converting any modern date into ethiopic.

However, it's not the case that extended year is the current era. Japanese and Chinese are the counter-examples.

We could certainly say that the extended year uses the current era for calendars that have a current era that is not expected to change.

[Manishearth]

However, it's not the case that extended year is the current era. Japanese and Chinese are the counter-examples.

I don't think I'm saying that at all?

I'm saying that extendedYear should be for an era that is actually valid for the calendar: having Amete Mihret be the epoch for both violates that premise. This is not a route that was proposed but it's worth addressing when we are discussing giving them the same era.

Having Amete Alem for both does not violate that premise, but has the problem that it uses an uncommon era.

Having them differ on era is also fine, though then they don't nicely match up, which is sad.

[sffc]

I see, so you're saying that there is a solid mental model behind both proposed solutions (Amete Alem for both, or Amete Alem for ethioaa and Amete Mihret for ethiopic) without actually taking a position on which you prefer. Is that right?

[sffc]

Having a framework for coming up with era names has helped us a lot.

Here's a proposal for a framework for choosing the extended year epoch:

1. Does the calendar have a current era which is expected to remain stable for the foreseeable future?
   • If yes, use that era's epoch for the extended year.
   • If no, continue.
2. Does the calendar have an alternative numbering scheme that is stable and widely understood?
   • If yes, use that numbering scheme for the extended year.
   • If no, panic! (I think all CLDR calendars hit one of the first two cases)

Applying this framework:

• Gregorian, Buddhist, ROC, Coptic: all hit case 1 trivially.
• Ethioaa: case 1, aa era.
• Ethipoic: case 1, am era.
• Japanese: case 2, ce era.
• Chinese: case 2, related iso numbering.

Thoughts?

[Manishearth]

I see, so you're saying that there is a solid mental model behind both proposed solutions (Amete Alem for both, or Amete Alem for ethioaa and Amete Mihret for ethiopic) without actually taking a position on which you prefer. Is that right?

Correct.

I have pros and cons for both as stated, but no clear preference.

[Manishearth]

That framework seems good to me.

[sffc]

@robertbastian Could you propose a framework that leads to aa as the extended year era for ethiopic?

[robertbastian]

If one calendar's eras are a superset of another calendar's eras, the primary era should be in the subset. This applies to both Ethiopic and Japanese.

[justingrant]

@sffc's proposed framework looks good to me.

leads to aa as the extended year era for ethiopic?

My understanding was that the main difference between ethiopic and ethioaa was that they used different extended year epochs. If they both used the same extended year numbering, then why have both calendars?

[sffc]

My understanding was that the main difference between ethiopic and ethioaa was that they used different extended year epochs. If they both used the same extended year numbering, then why have both calendars?

Extended year numbering is largely an internal detail. None of this impacts the formatting; only what happens when you get or set year. But, "largely internal details" need to be ironed out in the same way that weird edge cases of date differences need to be ironed out.

[ptomato]

We discussed this in the Temporal meeting 2025-04-03. Inheritance (as in the programming concept) is not a framing in which most people think of calendars, so we're skeptical about applying that as a framework.

[justingrant]

My overall feedback is that ethioaa should have an extended year epoch based on the AA era, and ethiopic should have an AM-based extended year epoch, because of the following assumptions and conclusions:

• The main reason that CLDR has long had two different Ethiopian calendars is that that in Ethiopia, the same real-world year can be represented by either an AA or an AM date.
• In other words, choice of preferred era is the main reason why a developer would choose one calendar or the other. If developers want to use AA dates, they'll use ethioaa (the "AA" is right there in the name!) and while if they want AM dates then they'll use ethiopic.
• Because era choice is the primary reason that there are two Ethiopian-related calendars, it will be least surprising for developers to have an extended year that matches their preferred choice of AM vs. AA era.

Am I correct in that the only calendar that would have different behavior between the two proposed frameworks is ethiopic ? If yes, then I'd vote for Shane's framework because it matches developer expectations more closely for ethiopic, and doesn't make things worse for other calendars.