Language codes

Lang Gérard gerard.lang at
Mon Sep 14 08:37:30 CEST 2009

Thank you.
You are entirely right concerning UK/GB..
Yes, I think it would be useful to correct "nine" to "ten" inside line six of section 2.4 of RFC 5645. 

-----Message d'origine-----
De : ietf-languages-bounces at [mailto:ietf-languages-bounces at] De la part de Doug Ewell
Envoyé : samedi 12 septembre 2009 21:25
À : ietf-languages at
Objet : Re: Language codes

"Lang Gérard" <gerard dot lang at insee dot fr> wrote:

> UK is one of the ten "exceptionally reserved" alpha-2 code elements 
> inside ISO 3166-1.
> This reservation has been asked by BSI, and accepted by ISO 3166/MA, 
> the reason being that in the initial cctld IANA  register that should 
> have contained ".gb", due to a personal error of John Postel, "uk" was 
> registered.

The United Kingdom, understandably, wanted to prevent the confusion that would have been caused by 'UK' being used for any other country or country-like entity, such as Ukraine.

As a side note, I believe there is a sharia-like law that provides for cutting out the tongue of anyone who claims publicly that Jon Postel ever made an error. :-)

> So I have no problem if the alpha-2 code elements FX andUK were added, 
> and the clause 2.4 New Region Subtags  of RFC 5645 accordingly 
> modified concerning FX and UK.

'FX' is already in the Language Subtag Registry, because it was a valid assigned code element in ISO 3166:1998, as referenced by RFC 1766.  This is explained in RFC 4645, Section 2, and in RFC 5645, Section 2.4.

Neither 'FX' nor any of the other exceptionally reserved ISO 3166-1 code elements exactly duplicates an existing assigned code element, except for 'UK'.  I argued against adding 'UK' to the Registry on the basis that it would create an exact, but non-preferred, duplicate entry. 
Adding new *preferred* duplicate entries updates best practice, such as in the pre-RFC 4646 days when the registered tag "art-lojban" was deprecated in favor of the ISO 639 code element 'jbo'.  Adding new
*non-preferred* duplicate entries just creates aliases (reducing
stability) and legitimizes non-preferred practice.

> By the way, following the insertion of SU inside the list (ISO 3166/MA 
> N 563  2008-06-27), there are currently ten exceptionally reserved
> alpha-2 code elements, and not nine as written on line six inside 
> clause 2.4 of RFC 5645.

You're right; that section was written before June 2008 and not corrected after the ISO 3166 status of 'SU' changed.  Feel free to file an erratum against RFC 5645 if you like.  However, this doesn't affect the Language Subtag Registry or any of the policies governing it, since 'SU' (like 'FX') is already in the Registry, having been a valid ISO
3166:1998 code element.

Doug Ewell  |  Thornton, Colorado, USA  | RFC 5645, 4645, UTN #14  |  ietf-languages @

Ietf-languages mailing list
Ietf-languages at

More information about the Ietf-languages mailing list