Hello,
In the upcoming JDK9, we are planning to modularize the locale data
available in the JDK for the JEP 200: The Modular JDK [1]. One of the
open issues of the JEP is this:
"The jdk.localedata module may eventually be split into multiple,
region-specific modules for added flexibility in configuration."
In JDK8, we include English locales into rt.jar, and the rest into
localedata.jar in the extension directory. This is OK but does not give
much flexibility on smaller environments, such as SE Embedded, where
they just use rt.jar and English locales are only available ones.
So here we are soliciting ideas on how to divide the locale data into
smaller entities, as the modular JDK would make it possible. Here are
some possible choices with pros/cons:
- Per language: Most straightforward choice, but it would create too
many modules.
- Per geological area: Smaller number of modules. Mapping geos into
locales may be complicated.
- Latin vs. other scripts.
- English vs. the rest: current implementation. Not flexible, but
practical/performant for the JDK full profile.
[1]: http://openjdk.java.net/jeps/200
Naoto
