On Fri, 1 May 2026 21:07:41 GMT, Naoto Sato <[email protected]> wrote:
>> Hmm, I was not talking about whether the CLDR data includes the input >> skeletons, but the adapter itself. >> >> As I understand it, the old code only ever constructed`LocaleResources` from >> a `ResourceBundleBasedAdapter` adapter. The new code introduces a fallback >> for the case when the CLDR adapter is not a `ResourceBundleBasedAdapter`. I >> am fine with it if it is intentional, I just noticed it differed from the >> original semantics. > > OK, now I got what you are saying. However, it is not possible that the CLDR > adapter is not a ResourceBundleBasedAdapter, as `CLDRLocaleProviderAdapter` > extends `JRELocaleProviderAdapter` which implements > `ResourceBundleBasedAdapter`. Exactly, I understand right now it isn't possible based on the current class definitions. It's just that the new code opens the possibility for it to accept such cases if that were to ever happen in the future (which again is unlikely) when the previous code did not. At face value, the code implies it is OK to have a non `RBBA` adapter there, when that isn't the case, so even a comment saying that isn't the case provides clarity. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/31006#discussion_r3175268089
