Re: [Ietf-languages] BCP47 violation in the recent extlang ajp change

[Doug Ewell <[e-mail]>]

Christian Despres wrote:

> In the recent IANA subtag registry update, the extlang "ajp" was
> deprecated, and its preferred value was changed from "ajp" to "apc".
> I believe this is in violation of the RFC5646 component of BCP47,
> specifically item (3) of
> <https://www.rfc-editor.org/rfc/rfc5646.html#section-2.2.2>, which
> states that the values of the subtag and preferred value fields of an
> extlang registry entry MUST be the same. This should be reverted.

Christian is right, and I will post forms shortly to correct this situation.

The same scenario occurred four years ago, when I noticed the violation of Section 2.2.2 in the deprecation of extlang subtags for ‘lsg’ (Lyons Sign Language), ‘rsi’ (Rennellese Sign Language), and ‘yds’ (Yiddish Sign Language). These subtags were deprecated in separate ISO 639-3 change cycles between 2015 and 2018. The story can be read at https://mailarchive.ietf.org/arch/msg/ietf-languages/_uxUBEppvHL1Xkq1JKWb2lpk_Fo/ .

The conflict is between Section 2.2.2, which requires extlangs to have a P-V that matches the subtag value (as Christian noted, and as I noted in 2019), and Section 3.1.7, which says that the P-V of a deprecated subtag should be updated as necessary to avoid “chaining” of subtags (‘xxx’ → ‘yyy’ → ‘zzz’). At that time, it was agreed (at least by silent assent) that Section 2.2.2 took precedence, and we need to do the same here, at the very least for consistency.

As far as I can tell, this situation has never occurred before with a spoken language (where the Preferred-Value and Prefix match), only with a sign language (where the Prefix is always “sgn”), nor has it ever occurred when the deprecated subtag represented a language that was merged into another, instead of declared non-existent (so that the primary language subtag has its P-V changed, instead of removed). The same rules for retaining the extlang’s P-V apply, though.

Note that this correction affects only the extlang, not the primary language subtag for ‘ajp’, which correctly had its P-V changed to “apc”.

> The way I believe this sort of situation is supposed to be handled in
> tag normalization is that a compliant processor will first convert the
> tag "ar-ajp" to simply "ajp", and then notice that the "ajp" is itself
> deprecated, at which point it will replace "ajp" with "apc". (The spec
> itself could probably be modified to permit an extlang's preferred
> value to differ from the value of the subtag itself without any issue,
> but that's beside the point).

In a perfect world, it would be best to be able to keep RFC 5646 and publish a short addendum clarifying how to resolve the conflict between sections 2.2.2 and 3.1.7.

In the IETF, however, I believe it would be necessary to publish a new RFC that replaces RFC 5646 in its entirety, all 84 pages. That would not only restart the whole long, painful RFC process that many of us hope to avoid, but would also open up all of RFC 5646 to being reevaluated and criticized. For example, there would likely be calls to replace or supplement ISO 639-3–based subtags with Glottolog, which would be a severe setback for stability and usability of BCP 47.

> Sorry if this is not the appropriate place to raise this concern.

It is exactly the right place. Thank you for bringing it here.

Issues like this are one of the main reasons we have a two-week review period for all change proposals, during which all list members are encouraged to check the proposals for all types of problems, including but not limited to BCP 47 violations.

--
Doug Ewell, CC, ALB | Lakewood, CO, US | ewellic.org