default charset for imlicit locale specificatio

[Andy Koppe]

2010/1/20 Corinna Vinschen
>> > I also noticed that on Linux two-letter settings like "de" or "ja" do not
>> > change the charset from ASCII to something else.
>>
>> Such locales don't usually exist on Linux, i.e. it's probably that
>> setlocale is failing, leaving the
>> program in "C".
>
> Nevertheless, maybe we should treat the languages w/o TERRITORY the same?

Not sure about that. If Linux did support them, they most likely
wouldn't use ASCII, since the point about the charset not actually
allowing the language to be written would apply again. Hence I'd go
with the language's charset, unless Windows doesn't actually allow to
obtain a territory-independent charset for a language, in which case
I'd go with UTF-8.

Andy