Re: [OT] Reusing the same property (was: RE: PRI #202: Extensions to NameAliases.txt for Unicode 6.1.0)

2011/8/31 Doug Ewell <doug_at_ewellic.org>:
> I don't know if this was what Philippe had in mind, but it reminded me
> of a situation in the world of language tagging.
>
> Apparently ISO 639-3/RA got a request, from an individual associated
> with a Very Well-Known Web Site, to change the 639-3 code element for
> the Wawa language from 'www' to something else.  Turns out that the site
> uses language codes in its URLs to link to different language versions:
>
> - www.example.org links to the generic site
> - en.example.org links to the English-language site
> - fr.example.org links to the French-language site
>
> And they wanted to have a site in Wawa, and encountered a name
> collision.

Here again, it's impossible to avoid collisions with all uses of
identifiers (names or numeric identifiers) that could occur within
specific applications. But at least ISO 639 provides a solution : you
can use a prefix that is unambiguously "private use" and use such
prefix to add the conflicting identifier. In ISO 639, the "q" prefix
will work well, so you can use "q-www" and still make use of the
standard ISO 639 prefix to avoid the collision with standardized host
names associated to web services offered in any domain.

The domain name system allows amny more solutions that will never
conflict with ISO 639 (you could as well use a digit for prefixing the
language code).
For this reason, there's absolutely no need for ISO 639 to provide
such aliases. There does exist standard aliases in ISO 639 (for
example several historic 2-letter codes in ISO 639-1, the technical
and bibliographic ISO 639-2 codes, 3-letter codes in ISO 639-2 and ISO
639-3) but their origin only comes from previous standards which
competed to the same application before ISO 639, and that were unified
in the same standard.

But there's no reason to pollute the unified namespace more (because
it would in fact create even more conflicts requiring an escaping
mechanism for some applications). But this also mean that any
well-behaved standard *unified* namespace has to include a private-use
space, as it allows integration and coexistence with other standards
or applications initially not designed to identify the same thing (for
example, here mixing the identification of a language, and the
identification and addressing of an Internet host)
Received on Wed Aug 31 2011 - 11:58:11 CDT