Doug Ewell doug at
Tue Jan 21 16:53:08 CET 2014

Mark Davis ☕ <mark at macchiato dot com> wrote:

> By "unstable", I mean that a tag valid in version V becomes invalid
> in V+1.

I think you're talking about the case where ISO 639-3/RA removes an
encompassed language from its macrolanguage. However, I believe we would
still have stability in that case.

Imagine that the RA decided to remove Jakun ('jak') as a member of the
macrolanguage Malay ('msa', or as we call it, 'ms'). In the Registry,
the Macrolanguage field for primary and extended language subtags 'jak'
would be removed, since that field tracks 639-3 exactly. But both
subtags would still remain, and the extlang would still have a Prefix
and Preferred-Value of 'ms' (the Prefix MUST NOT be removed -- 3.4). The
extlang would probably be deprecated, but that does not affect validity

Even if 639-3 withdrew the code element 'jak' entirely -- say, if Jakun
were determined to be a dialect of Malay, as Ethnologue suggests -- both
primary and extlang subtags would be deprecated but still valid.

Thus the extlang relationship doesn't really present any more
instability than that of primary language subtags, and thus the validity
rules didn't really have to allow mismatched language-extlang pairs to
defend against instability.

Did we discuss either of these scenarios in LTRU?

Doug Ewell | Thornton, CO, USA | @DougEwell ­

More information about the Ietf-languages mailing list