era and year in ICU calendars

Frank Yung-Fong Tang

ftang@google.com

2022-09-06

Calendar

GregorianCalendar

(“gregorian” / “gregory”)

Public Headers

Private Headers / Internal

TaiwanCalendar

(“roc”)

JapaneseCalendar

(“japanese”)

BuddhistCalendar

(“buddhist”)

PersianCalendar

(“persian”)

IndianCalendar

(“indian”)

IslamicCalendar

​

​

​

​

HebrewCalendar

(“hebrew”)

CECalendar

CopticCalendar

(“coptic”)

EthiopicCalendar

​

​

​

​

ChineseCalendar

(“chinese”)

DangiCalendar

(“dangi”)

Interface

Abstract Class

Concrete Class

(“key” / key in JS)

Extended Class Diagram of Calendar Classes inside ICU

(“ethiopic”)

(“ethiopic-amete-alem” / “ethioaa”)

(“key”)

Typed objects:

(“islamic”)

(“islamic-civil”)

(“islamic-umalqura”)

(“islamic-tbla”)

(“islamic-rgsa”)

(“iso8601”)

Five Different Models for year and era

I

II

IV

-2 -1 0 1 2

3 2 1 1 2

-2 -1 0 1 2

1 2

1 2

1 2

1 2

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

V

III

-2 -1 0 1 2 5499 5500 1 2

Five Different Models for year and era

-2 -1 0 1 2

3 2 1 1 2

-2 -1 0 1 2

1 2

1 2

1 2

1 2

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

One Directional Line (e.g. “iso8601”, “buddhist”, “hebrew”, “islamic”, "ethiopic-amete-alem”, etc)

Irregular Length Directional Lines (e.g. “japanese”)

Two Directional Lines Opposite Direction (e.g. “gregory”, “roc”, “copic”,)

Fixed Length Directional Lines (e.g. “chinese”, “dangi”)

Two Directional Lines Same Direction ("ethiopic")

-2 -1 0 1 2 5499 5500 1 2

I

II

IV

V

III

Calendars in ICU 71.1

• “iso8601”
• “persian”
• “indian”
• “hebrew”
• “islamic”
• “islamic-tbla”
• “islamic-civil”
• “islamic-umalqura”
• “islamic-rgsa”
• “ethiopic-amete-alem”
• “buddhist”

• “ethiopic”

• “gregory”
• “coptic”
• “roc”

• “japanese”

• “chinese”
• “dangi”

I

II

III

IV

V

Type I - One Directional Line

-2 -1 0 1 2

Calendars in ICU:

• “iso8601”
• “persian”
• “indian”
• “hebrew”
• “islamic”
• “islamic-tbla”
• “islamic-civil”
• “islamic-umalqura”
• “islamic-rgsa”
• "ethiopic-amete-alem”
• “buddhist”

Since there is only one era, there are no need to specify “era”

I

Type II - Two Directional Lines Opposite Direction

3 2 1 1 2 1970 2022

Calendars in ICU:

• “gregory”: BC and AD
• “coptic”: BCE and CE
• “roc”: BEFORE_MINGUO and MINGUO
• Shift 1912 years

Should we define one single set of two values for all of them?

• “beforeCurrentEra” and “currentEra”

​

​

gregory: BC AD

​

3 2 1 1 2 110

roc: Before MinGuo MinGuo

3 2 1 1 2 1970 2022

coptic: BCE (Before Coptic Era) CE (Coptic Era)

​

3 2 1 1 2 1739

II

Type III - Two Directional Lines Same Direction

​

Amete Alem Amete Mihret

-2 -1 0 1 2 5499 5500 1 2 2014

Calendars in ICU:

• "ethiopic”
• AMETE_ALEM and AMETE_MIHRET

​

It could be viewed as a special case of Type IV

​

III

"ethiopic”

Side Note: “ethiopic” vs "ethiopic-amete-alem” calendar

​

​

​

Amete Alem Amete Mihret

-2 -1 0 1 2 5499 5500 1 2 2014

III

"ethiopic”

-2 -1 0 1 2 5499 5500 5501 7514

I

Amete Alem

"ethiopic-amete-alem”

• Amete Alem (ዓመተ ዓለም) = “The year of the world”
• Amete Mihret (ዓመተ ምሕረት) = “The year of mercy”

Ref https://ethiopiancalendar.wordpress.com/history/

Type IV - Irregular Length Directional Lines

​

-2 -1 0 1 2

1 2

1 2

1 2

1 2

Calendars in ICU:

• "japanese”
• One (or more) for each Japanese Emperor Ruling Period
• Length is irregular and unpredictable for the future

IV

Type V - Fixed Length Directional Lines

​

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

Calendars in ICU:

• “chinese”:
• “dangi”

​

It is really not a true era/eraYear in the calendar, but rather a hacky way to put 60 years cycle (0..59) with a cycle sequence into year and era fields.

​

era = extended_year / 60 => cycle sequence

year = exteneded_year % 60 => year in the cycle

​

People know the “year in the cycle” in the culture

No people know about the cycle sequence

​

V

Five Different Models for year and era

I

II

IV

-2 -1 0 1 2

3 2 1 1 2

-2 -1 0 1 2

1 2

1 2

1 2

1 2

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

0 1 …. 59

V

III

-2 -1 0 1 2 5499 5500 1 2

Calendars in ICU 71.1

• “iso8601”
• “persian”
• “indian”
• “hebrew”
• “islamic”
• “islamic-tbla”
• “islamic-civil”
• “islamic-umalqura”
• “islamic-rgsa”
• "ethiopic-amete-alem”
• “buddhist”

• “ethiopic”

• “gregory”
• “coptic”
• “roc”

• “japanese”

• “chinese”
• “dangi”

I

II

III

IV

V

Proposal to ECMA402 Intl-Temporal

• Type I: Do NOT use era and eraYear (return both as undefined)
• Type II (“gregory”, “coptic”, “roc”) : Two possible approaches
• Approach A: Define one single set of two values
• “currentEra” and “beforeCurrentEra” (could be other strings)
• Can NOT be “commonEra” “CE” or “beforeCommonEra” �“BCE”
• Approach B: Define one set of two string for each calendar
• Type III: (“ethiopic”): TBD
• Type IV: (“japanese”): TBD
• Type V: Not supporting the “year in cycle” + “cycle sequence”
• Treat “year in cycle” + 60 * “cycle sequence” as Type I.

In ICU4X

“gregory” era:

• “bce”
• “ce”

“japanese” era:

• “bce”
• “ce”
• "meiji"
• "taisho"
• "showa"
• "heisei"
• "reiwa"

“buddhist” era

• “be”

“indian” era:

• “saka”

“coptic” era:

• “bd”
• “ad”

“ethopic” era:

• “mundi”
• “pre-incar”
• “incar”

Why do we need to pass in an era code which can only be one value and return error if not?

Different semantic than ICU (see Type III)