[threeten-dev] Islamic calendar variants need disambiguation

[yoshito_umaoka at us.ibm.com]

> Yoshito-san,
> 
> I would like to get your opinion about identifying Islamic calendar 
> variants in CLDR. It defines "islamic" and "islamicc" in calendar.xml 
> for the astronomical and algorithmic variants respectively, while 
> JSR-310 needs to clearly identify the variants and distinctively 
> identify each of them, in order to identify the dates correctly.
> 
> For example, the Umm Al-Qura calendar and Microsoft's Islamic calendar 
> implementation with the Kuwaiti algorithm are both algorithmic but they 
> are different, so the date information is not interchangeable under 
> "islamicc" between different variants. A date in one variant may 
> correspond to a different date in another variant. So, in terms of 
> ensuring correct exchange of Islamic date information, not identifying 
> the precise variants in the CLDR name space falls short of the use case 
> expectation. In JSR-310, we anticipate Islamic end users might have 
> their preferred Islamic calendar variant set in their user profile, or 
> an application may be supporting different variants in producing some 
> output with Islamic date information.
> 
> May I ask if there is a chance to add more Islamic calendar variants or 
> enhance the calendar naming scheme to resolve this issue? Or is this 
> beyond the scope of CLDR? Do you suggest taking this to the CLDR list?

I understand the issue. I don't think it's beyond the scope of CLDR.
I think the major question is - how many variants are enough.

Historically, CLDR project was spawned from ICU project. ICU calendar
implementation is derived from a book - Caledrical Calculations
(Nachum Deshowitz, Edward M. Reingold - ISBN 0-521-56413-1).
In this book, "civil" calendar (arithmetic or tabular) is explained
along with "religious" calendar which is based on astronomical
observation. The "civil" algorithm explained in this book uses leap
year [2,5,7,10,13,16,18,21,24,26,29] that is different from Microsoft's
"Kawait algorithm".

I was looking at some web pages. This page
[http://www.staff.science.uu.nl/~gent0113/islam/islam_tabcal.htm]
illustrates at least four different arithmetical variations. It also
mentions that there are two possible variations for each intercalary
scheme based on epochs.

Also, by the nature of "religious" calendar, which may vary depending
on astronomical observation of regional authorities, it is obviously
not suited for identify the dates across different regions.

Here is what I think -

- CLDR type "islamic" is used for religious calendar now. Although
  actual date may vary, it is probably not worth adding variations.

- CLDR type "islamicc" is used for arithmetical calendar now. This
  type could  be separated into multiple different variations.

I'm wondering if you can create a CLDR ticket with proposed types?
I can add my comment and bring this to a CLDR meeting.

Thanks,
Yoshito