What a Locale is.... (Re: [Fwd]: Response to Mark's message])

[Harald Tveit Alvestrand]

--On onsdag, april 09, 2003 14:57:32 -0400 Martin Duerst <duerst at w3.org> 
wrote:

> At 10:08 03/04/09 -0700, Mark Davis wrote:
>> There is a misunderstanding here. However one defines locales, one needs
>> a language ID to be useful. For example, someone might define a locale to
>> include:
>>
>> written language ID
>> country ID of residence
>> country ID of citizenship
>> country ID of bank account
>> timezone ID
>>
>> [This is actually the defintion needed by a customer I was talking to
>> just yesterday.]
>
> This is a good example, in that it shows that different applications
> have very differing needs. I never would want a generally used
> locale definition to include the above items, because in particular
> citizenship and bank account have clear privacy implications.
>
> So I think we need to make it easier to write good applications,
> and continuing to wonder what a 'locale' might not be the best
> way to do that.

it seems to me that this example shows the fundamentally problematic 
quality of a locale: That the standardizers never got around to figuring 
out what the locale is a property *of*.

For language, it's pretty clearly the property of a piece of text or other 
material for transferring speech (in its widest sense), such as audio, 
video or pictures. By extension, it is used to specify desired properties 
of pieces of text (such as in user preferences).

For locale, as used in the POSIX family of standards, it is a multifacteted 
thing that is a property of the execution environment of a program.
But that doesn't make it a property of the program.
Nor is it a property of the user using the program (albeit possibly 
related).
Nor is it a property of the data the program is processing.
Nor is it a property of the location the program is executing in.
Nor is it a property of the location the user is resident in.

I'm tempted to suggest a radical conclusion:

The "locale" concept is unsuitable for reference in communications or data 
storage, and the word should therefore not be used on this mailing list.

Instead of referring to "locale", all participants should specify *what 
they want to do differently* - ie that in a certain context, they need to 
specify dates in one of seven formats, and the formats need to have names.

And so on for other cultural elements - currencies HAVE names, languages 
HAVE names, countries HAVE names.

(personally, I want the nn/nn/nn date format to have the name "useless" - 
but that's unlikely to win the day :-)

                          Harald