> > > Any proposal to assign variants on a systematic, formulaic basis > -- not > > three or four, but dozens or hundreds or thousands -- ought to > make us > > stop and wonder why we don't establish an extension for the > purpose > > instead: > > The point of creating an extension is to provide additional > orthogonal > information that is packaged up with the language tag. (IMHO this > is > rather silly, as few databases allow only one field, but who knows.) > They aren't meant to be used for specification that of language > variant > information; that is semantically part of the language tag and > belongs > in the variant field. > Well, that isn't necessarily so. I know that the RFC-to-be says that extensions are "orthogonal". But an extension could be created that wasn’t entirely orthogonal to language identification. It might provide finer-grained identification (such as 639-6 is purported to provide) than variants might provide. I think that having an extension proposal would actually be the test of whether it were an appropriate use or not. Just as having 639-6 in a relatively mature state is an important pre-condition to its potential incorporation into the corpus of BCP 47. Addison
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.