[loc-en-dev] Recommendation for language tags in old Locale

[Yoshito Umaoka]

Eike Rathke wrote:
> Hi,
>
> I browsed http://sites.google.com/site/openjdklocale/ and this mailing
> list, maybe I overlooked it. Are there recommendations for how to stuff
> a language tag into the old Locale using 3 fields Language/Country/Variant?
>
> e.g., I could imagine that a script belongs more to a language than
> a country, thus sr-Latn-RS could be transported as 'sr_Latn' in
> Language, and 'RS' in Country as usual. What about dialects? Should they
> go into Variant?
>
> Thanks
>   Eike
>   
In the past, some folks try to distinguish a language from a locale and 
some other folks thinks it is not worth making clear distinction.  We do 
not have any agreed consensus yet.  But, not matter you think language 
tag cannot be used for cultural preferences (locale), it can be used for 
decide the default cultural preference set (for example, see w3c i18n 
article - http://www.w3.org/International/questions/qa-accept-lang-locales)

The language tag - "sr-Latn-RS" - indicate Serbian written in Latin 
script used by Serbia and Montenegro.  If we try to interpret this as a 
locale id, it means cultural preferences using Serbian written in Latin 
used by Serbia and Montenegro for localizable names AND cultural 
conventions (such as calendar system, currency, measurement units...) 
commonly used in Serbia and Montenegro.  Java Locale is an identifier 
for locale.  So "sr_Latn" and "sr_Latn_RS" could result different 
behaviors in some Java locale services.

In BCP47, if you need to distinguish a dialect from another, registering 
the dialect as a variant would be the right approach.  In Locale, it is 
mapped to variant field.

-Yoshito