an old item: es-americas

[Doug Ewell]

Addison Phillips [wM] <aphillips at webmethods dot com> wrote:

> If we were to permit the use of any M49 code for which there currently
> is no ISO3166 code, we might have the situation in which the 3166 code
> comes later (okay, that's unlikely). Similarly, we don't want some
> people to use the 3166 and some to use the M49 for the same thing. So
> the change in rule that we are contemplating is:
>
> a) use the iso3166 alpha2 code dating from January 2003 or any new
> unambiguous assignments (that do not overlap with the codes assigned
> or transitional on that date).
> b) if an ambiguous (re)assignment is made, then use the corresponding
> M49 code (assigned via registration)

Not all that unlikely.  M49 includes codes for the Channel Islands and
Isle of Man, which 3166 does not.  So clearly the only choice today, if
for some reason you need to specify English as spoken in Guernsey, is
en-830.  But if 3166 can add a code for Åland Islands, they could
certainly add one for the Channel Islands, or Guernsey specifically, in
the future.

You are already considering a "January 2003" rule that differentiates
the treatment of the CS reassignment from that of other reassignments in
the past.  Just apply a similar "April 2004" rule to the use of M49
codes, whereby any country-like entity that has no 3166 code as of April
2004 (thus excluding Åland Islands) must be denoted by the M49 code,
EVEN IF a 3166 code is subsequently assigned -- in which case it may not
be used in RFC3066bis language tags, just as CS (meaning Serbia and
Montenegro) may not.

This way, you could build strictly limited use of M49 codes into the
RFC, avoiding the need to burden the registration process with them.

> The question is whether to add another rule. The variations we could
> add would be:
>
> - if no iso3166 code exists, you may use the M49 code (that is,
> permitting regional and sub-regional codes)
> - specifically reference the regional (supra-national) codes to allow
> them (that is, separately reference these specifically)
>
> Here are some of the potential codes:
>
>   es-019 (America)
>   es-419 (Latin America & Carribean)
>   es-013 (Central America)
>   es-005 (South America)
>   es-021 (North America)

I think using an existing coding system is self-evidently better than
encouraging individuals to propose vague identifiers like "americas."
Registering variants should really be a last resort.

John Cowan <jcowan at reutershealth dot com> wrote:

>> - if no iso3166 code exists, you may use the M49 code (that is,
>> permitting regional and sub-regional codes)
>
> I would propose a slight reformulation of the first alternative:
>
> Any M49 code that does not refer to a country may be used.

The "geographic" codes should all be permissible, but I would argue
strongly against allowing the "economic" M49 codes.  A category like
"least developed countries" has no place in language tagging.

-Doug Ewell
 Fullerton, California
 http://users.adelphia.net/~dewell/