<i18n dev>  RFR: 8181157: CLDR Timezone name fallback implementation
xueming.shen at oracle.com
Mon Apr 23 20:06:50 UTC 2018
Here some comments
Ln#58: to use Stream.toArray(String::new) ? no sure which one is faster
is it worth considering to check all possible empty slots in "names" here
(from index_std_long to index_std_short?) to avoid check/load "compact" multiple
times? it appears doable for deriveFallbackNames() but I'm not sure about
the invocation at #ln100, so a "?".
Ln#674: (not in updated code) why update the "local" jreMetaMap?
just wonder it'd better to use Optional.ifPresentOrElse(...) here
Ln:707: (don't know which one is better though :-))
just wanted to confirm the "locale.resources.debug" is something we wanted to
be a public interface or not?
On 04/17/2018 02:28 PM, Naoto Sato wrote:
> Please review the changes to the following issue:
> The proposed fix is located at:
> This RFE is to implement CLDR time zone names fallback mechanism . Prior to this fix, time zone names are substituted with English names. With this change, it is generated according to the logic defined in TR 35. Here are the highlight of the changes:
> - CLDRConverter has changed to not substitute/invent time zone display
> names, except English bundle for compatibility & runtime performance.
> - For English bundle, copy over COMPAT's names as before.
> - CLDRConverer now parses regionFormat/gmtZeroFormat/exemplarCity
> CLDR provider:
> - CLDRTimeZoneNameProviderImpl is introduced to generate fallback names
> at runtime.
> - If COMPAT provider is present, use it also as a fallback, before the
> last resort (GMT offset format)
>  http://www.unicode.org/reports/tr35/tr35-dates.html#Time_Zone_Names
More information about the core-libs-dev