On Wed, 13 Jul 2022 16:54:48 GMT, Naoto Sato <na...@openjdk.org> wrote:
>> I would like to emphasize that this issue is complicated. Plural rules vary >> from language to language, adding plural strings to resource bundles adds >> costs in terms of translation and testing. >> >> If we were to support proper plurals, one may use ChoiceFormat, though it >> seems to fail for Polish, see >> https://stuartgunter.wordpress.com/2011/08/09/java-i18n-pluralisation-using-choiceformat/ >> >> Another solution is to pull in ICU4J library, see >> https://stuartgunter.wordpress.com/2011/08/14/even-better-java-i18n-pluralisation-using-icu4j/ >> >> Yet another solution is to drop the words altogether and simply show a >> number (of bytes in this case). Alternatively, use a common suffix such as >> "B", but this may not look standard. > > True plurals (as in LDML) is not yet fully supported in the JDK. > (`CompactNumberFormat` does for its own purpose > (https://bugs.openjdk.org/browse/JDK-8222756), but not for general use) For > this case, I think `ChoiceFormat` is the mid-ground solution, as Swing is > only localized in a handful of languages. another suggestion: to avoid locale-dependent complexity simply show the number of bytes as a number if length < 1000. this will also avoid involving the group separator - as far as I know, no language on earth uses group separator for numbers in the range 0...999 0 - "0" 1 - "1" 999 - "999" 1000 - "1 KB" etc. ------------- PR: https://git.openjdk.org/jdk/pull/9327
