Hi Thejasvi, Thanks for fixing this.
Since those new test cases depend on the CLDR localization, which might change in other implementations, those test cases should be in jdk/java/time/test directory, as "tck" tests should only test the spec. Please create a new test case for this in the "test" directory (with @modules jdk.localedata directive) similar to the existing TCK one. Also the current test in the TCK should enforce that it runs in Locale.US so that the result should match "GMT."
Naoto On 6/28/19 5:59 AM, Thejasvi Voniadka wrote:
Hi, Request you to please review this change. JBS: https://bugs.openjdk.java.net/browse/JDK-8154520 Description: At present, the "DateTimeFormatterBuilder.appendLocalizedOffset()" method formulates the base string as "GMT", without accounting for locale-specific transformations. This change is to return the localized version of "GMT" instead. So for example, instead of returning "GMT +5.30", it may now return "XXXX +5.30" where "XXXX" is the localized string for "GMT" for the locale associated with the formatter. I have used DateTimeTextProvider.getLocalizedResource() method to return the "gmtZeroFormat" value from CLDR/LDML corresponding to the given locale. The code defaults to "GMT" in the absence of such a localized value. Webrev: http://cr.openjdk.java.net/~vagarwal/8154520/webrev.1/ Additional notes: I preferred to update and reuse an existing test instead of creating a new one. It already has the niceties in place, and creating another method would mean some amount of code redundancy. However, if that's the recommended norm, then I can change it.
