[threeten-dev] CLDR calendar variants

[Dan Chiba]

Ok I understand your desire to address all issues before publishing the 
new "cv" scheme.

Imho, the issue is that "mshijri" is not specific enough to identify a 
variant. It seems there are multiple revisions with slight differences 
so giving them a single name would confuse them and defeat the purpose 
of assigning a "cv" value. Even if the presence of multiple revisions 
was not true, it does allow adjusting the epoch and that alone fails the 
stability criteria that each registered variant must satisfy. I think 
all registered variants must be completely stable, meaning that the 
definition of the dates can never change and there should be no 
configurable element. If there is one, like the epoch of "mshijri", each 
possible configurable value should be registered as a distinct variant 
instead. (Or, yet another level in the mechanism; e.g. "ce" for the 
epoch --- u-ca-islamic-cv-mshijri-ce-xxxxx. This sounds too much and 
inadequate.)

As for "caltab" it is my understanding that it is clearly specified and 
there is no stability issues as "mshijri" has.

Overall, I think it is reasonable to require a certain level of clarity 
and stability before a variant can be registered. Would it be possible 
to request help to someone from Microsoft in the CLDR community for 
specifically identifying the "mshijri" variants? Then it should be 
possible to find good "cv" values for them.

Regards,
-Dan

On 2/25/2013 2:58 PM, yoshito_umaoka at us.ibm.com wrote:
> > Because 310 has implemented and counts on the identification with "cv",
> > we would like it to earn an official status, much more than we like to
> > sort out the question around one of the variants we considered
> > supporting. It is discouraging to see the base "cv" mechanism and 
> stable
> > variants delay due to the relatively minor question raised. Could it be
> > considered to detach this issue around Microsoft Hijri calendar, so the
> > rest could avoid deferral?
>
> We'd like to publish the new stuffs after resolving all issues.
> Unfortunately, CLDR project members are now finalizing CLDR 23 data - 
> and data changes are already closed except ship stopper fixes.
>
> However, we've already concluded that we need "cv" and I don't expect 
> we have another discussion about this. Only the remaining issue is 
> type values and its definition. More specifically, as I mentioned in 
> the previous message, "caltab" and "mshijri".
>
> -Yoshito
>