You are right. Those could happen in theory. As far as I checked, there
are two eras which are shorter than one year, 朱鳥 in 686 (Julius) and
天平感宝 in 749 (Julius). Both have only a few months. So I wouldn't say
it's quite a risk, especially under the current rule, one era per emperor.
One issue is that the current Japanese calendar implementation isn't
ready for such extreme cases, like a one-day year which breaks some
assumptions. Currently, java.time uses the j.u.JapaneseImperialCalendar
implementation. So java.time has the same restrictions. You may want to
implement the perfect Japanese calendar which supports all the past eras
(with all the different transition rules) and all the possible future
era transitions.
Masayoshi
On 7/22/2014 10:33 PM, Stephen Colebourne wrote:
A Japanese Emperor might only reign for one day or one week, which is
far less time than a JDK update. It seems to me that one additional
era is quite a risk to be baking into the JDK.
Stephen
On 22 July 2014 09:05, Masayoshi Okutsu <masayoshi.oku...@oracle.com> wrote:
The assumption here is that it's enough to support one additional era until
the era is added to the built-in eras in update releases.
Masayoshi
On 7/22/2014 4:52 PM, Stephen Colebourne wrote:
It seems that the system property only supports one additional era.
Given that Japanese eras are based on human lifetimes, that seems an
invalid assumption.
Stephen
On 22 July 2014 08:06, Masayoshi Okutsu <masayoshi.oku...@oracle.com>
wrote:
Hello,
Please review the change for JDK-8048123. This change removes all the era
definitions in ${java.home}/lib/calendars.properties. (The property file
will eventually be gone later.) The major change is that any new era of
the
Japanese calendar should now be defined using new property
"jdk.calendar.japanese.supplemental.era".
https://bugs.openjdk.java.net/browse/JDK-8048123
Webrev includes some unrelated cleanups.
http://cr.openjdk.java.net/~okutsu/9/8048123/webrev.00/
Thanks,
Masayoshi
