[threeten-dev] Julian and Ethiopian calendar system

[yoshito_umaoka at us.ibm.com]

> >> Hi Yoshito,
> >> Does LDML have any "ca" code for the Julian calendar system? I cannot
> >> see one in 
http://unicode.org/repos/cldr/trunk/common/bcp47/calendar.xml
> >> I'd like to add a class representing the Julian calendar system at
> >> some point, and it should really have a "ca" code, hopefully 
"julian".
> >> It would only need localization for the calendar system name, as the
> >> other text (months/day-of-week/era) would be the same as the 
"gregory"
> >> calendar.
> >> (Note that I presume the "gregory" calendar is proleptic, applying 
the
> >> same rules back into far history)
> >
> > CLDR was originally spawned from ICU project long time ago. It's 
independent
> > from ICU project now, so it is not necessarily follow the original ICU
> > model. In ICU, Gregorian calendar is not always proleptic. By default, 
it
> > uses 1582-10-15 as the switch date between Julian and Gregorian. The
> > implementation was contributed to JDK - therefore,
> > java.util.GregorianCalendar also use the same logic (see
> > GregorianCalendar#setGregorianChange).
> 
> OK, so if you intend "gregory" to mean "the Julian-Gregorian calendar
> with a cutover in 1582", thats fine by me. But we are going to need a
> proleptic Julian calendar, which uses Julian rules throughout all
> time. I'm not convinced thats a variant as it results in dates that
> are much more different than the variants of the Hijrah.

I don't think LDML/CLDR intend "gregory" to mean the Julian-Gregorian 
calendar. I think it's just ICU/JDK implementation.

For "formatting date", Gregorian and Julian do not require different set 
of symbol definitions. They just use slightly different rule for 
calculating dates. From this point of view, I think they are nothing much 
different from what we did for Islamic calendar variants.

> > Strictly speaking, ICU/JDK GregorianCalendar might be called 
WesternCalendar
> > or something - because it's hybrid. I'm not sure there is convincing 
reason
> > to add "Julian". This might also fall into calendar "variant" 
category.
> >>
> >> Has any thought been given to converting the two ethiopian calendar
> >> systems to be "cv" variants? "ethioaa" and "ethiopic" are variants of
> >> the same calendar system AFAICT.
> >
> > Good point. It might make sense. Although they require different 
symbols for
> > era (has impacts to format symbol data unlike Islamic variants). 
Because we
> > invented "cv" for supporting Islamic requirements recently, we did not 
have
> > chance (or urgent needs) to see if this concept is applicable to 
others yet.
> > I'll bring these topics to CLDR TC, but I assume these are not urgent. 
For
> > now, CLDR project folks are busy on shipping CLDR 23 - so this is 
something
> > we would consider after CLDR 23.
> 
> It looks like we don't have Ethiopic it for JDK 1.8, but feel it would
> be good to decide on whether this is a variant, and whether that
> impacts the design of variants in LDML.
> 

OK.

-Yoshito