Search code examples
javalocale

Difference in initialization of a Locale object

Is there any difference between this two ways of creating a Locale :

Locale uk    = Locale.UK;

Locale en_uk = new Locale("en", "UK");

The only difference I have noticed so far is that getDisplayCountry returns different values. The first United Kingdom and the second UK , which does not bother me very much. Is there besides that any difference? Especially because I would like to adopt the first style for other regions like Germany, France, Korea etc

Solution

  • tl;dr

    • Wrong country code.
    • Do not use Locale constants. Use Locale.of.
    Locale.of( "en" , "GB" )  // For English language, and cultural norms of United Kingdom.

    Other examples:

    • Locale.of( "en" , "CA" ) for English in Canada.
    • Locale.of( "fr" , "CA" ) for French in Canada.
    • Locale.of( "en" , "US" ) for English in United States.

    GB, not UK

    For your purpose, the correct country code for the United Kingdom of Great Britain and Northern Ireland is GB, not UK, as commented by Mark Rotteveel.

    CLDR

    Modern Java (9+) uses by default the Common Locale Data Repository (CLDR) as the primary source of localization data. Entry # 827 in the CLDR shows a country code of GB for United Kingdom.

    Standard country codes

    The Javadoc for Locale states that the country/region code must be either:

    If curious, dump to console:

    Arrays.stream( Locale.getISOLanguages() ).forEach( System.out::println );  // All 2-letter country codes defined in ISO 3166.
Arrays.stream( Locale.getISOCountries() ).forEach( System.out::println );  // All 2-letter language codes defined in ISO 639.

    The Javadoc also notes that a complete list can be found in the IANA Language Subtag Registry with a tip to search for Type: region. There we find this entry telling us to use GB for the United Kingdom.

    %%
Type: region
Subtag: GB
Description: United Kingdom
Added: 2005-10-16
Comments: as of 2006-03-29 GB no longer includes the Channel Islands and
  Isle of Man; see GG, JE, IM
%%

    Validator

    The Comment by andrewJames points us to this online validator by Christoph Schneegans for IETF BCP 47 language tag with language and country code combined.

    There we see en-UK fails, while en-GB passes.

    Avoid Locale constants

    You should ignore those constants defined on Locale class such as Locale.UK and Locale.CANADA_FRENCH. Their creation was a hasty poor decision in design. For more info, see my longer Answer.

    Instead, in modern Java (19+), use the convenient static factory methods Locale.of.

    Locale.UK, Locale.of( "en" , "GB" )

    See Locale.of in action.

    Locale locale = Locale.of( "en" , "GB" ) ;