Right: I agree completely.
I would point out, though, that in Web language negotiation (where you
have next-to-no information as it is), a web server that doesn't match
'nb' will not know to try 'no' or 'nn'--instead it'll go to the next
language in the Accept-Language stack (if there is one) or to default
(which is probably the language of the server/URL's owner).
This is a non-trivial problem. Especially since the server then
extrapolates the locale from what is ostensibly only the *language*
request (these things look like locale names for a reason).
I brought up the Chinese example because the generally accepted
"solution" is that zh_TW == Traditional and zh_CN == Simplified--and this
leaves other Chinese locales (zh_HK, zh_SG, not to mention the default
Chinese 'zh') in some kind of limbo. Yes, you can tell from the characters
which writing system you're using but when you're getting an http request
you won't have any characters to work with!
I have short topic and some source code on the website trying to depict
why this is not a simple problem: it's easy to parse the header and hard
to know what it means.
(http://www.inter-locale.com/demos/langnego.xtp)
thanks,
Addison
On Thu, 31 Aug 2000 Marco.Cimarosti@icl.com wrote:
> Addison P. Phillips wrote:
> > Differences in writing systems are much more problematic than the
> > Norwegian example. The Simplified/Traditional Chinese thing
> > leaps to mind, of course, [...]
>
> Right. I just notice that, in Unicode, this is not a display difference but
> an encoding one: corresponding simplified and traditional characters have
> different code points.
>
> > I imagine that the nb/nn codes will eventually predominate...
> > except that this means many Norwegian locale structures are
> > going to be messy: Java, for example, will default from nb
> > to default, not from nb to nn-if-it-exists or from nb to no.
>
> "Cascading" language selection is a great thing, and we are probably moving
> in that direction.
>
> But it is probably pointless to standardize the hierarchy of languages,
> because it is too much bound to the culture and preferences of each
> individual.
>
> E.g. you cannot assume that Italian is the preferred language for all
> Italians and, if not available, English is always the next best choice. The
> mother tongue of many Italian citizens is German (or French, or Catalan, or
> Albanian) and, although English is the most widespread foreign language,
> many Italians are more fluent in French, German, Spanish, Russian...
>
> Similarly, wouldn't it be plausible that some Norse people, in the absence
> of a Nynorsk interface, prefer a foreign (but familiar) language like
> English, rather than the domestic (but maybe not very well known) Bokmål?
>
> So, probably, the hierarchical list of language preferences should be a
> customizable part of the locale, as it is in fact in some environments.
>
> _ Marco
>
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:13 EDT