RE: [Ltru] Obsolete descriptions (was: "X" vs. 'X (macrolanguage)")

> From: Frank Ellermann [mailto:nobody@xyzzy.claranet.de]

> >> 3066 and 4646 went to great lengths that this is
> >> not possible, including a promise from ISO in the
> >> text.
>
> > Whate are you referring to? I am not aware that
> > IETF requested any promises from ISO wrt descriptions.
>
> Bottom of <http://tools.ietf.org/html/rfc4646#page-9>:
>
> The promise in RFC 3066 and 4646 is about alpha-2
> and alpha-3 codes, for an existing alpha-3 code
> for a language (identified by its description)
> there won't be a new alpha-2 code (with the same
> description because it's the same language).

Right. That has nothing whatsoever to do with descriptions.

> > SIL's pages in relation to this entry in ISO 639-3
> > are not wrong wrt ISO 639.
>
> Then the 4646 registry must be wrong, it's supposed
> to reflect whatever description ISO 639-1 offers for
> an alpha-2 code.  Caveat, I looked up the ISO 639-1
> "sw" on the SIL Web form, not some ISO 639-3 "sw".

There is no 639-3 "sw".

The site for 639-3 lists reference names as they exist in 639-3. The 639-2 site lists names based mainly on names as they existed in 639-1 and 639-2 prior to 639-3. (Some changes have been made since 639-3 was published) For 639-2, there isn't the same demand on the RA to add qualifiers to the names they record because there is no naming conflict in 639-2. (In cases where there is a naming conflict in 639-2, qualifiers have always been added.)

In principle, it's reasonable to expect the names for 639 would apply consistently across all parts. In practice, the 639-2 RA has been reluctant to make changes that would not be meaningful for their audience until they see some reasonable mitigation.

At this point (i.e. 4646bis era), I see no reason why the LSR could not rely solely on data provided by the 639-3 RA.

Peter