Re: rfc4646. Some analysis code

Chris Lilley wrote:

> Does that mean (since cel-gaulish was not on your irregular list, and also
> given that cel is a registered tag and gaulish is five-to-eight letters)
> that I can now write things like

> cel-GRK-gaulish

> to describe, say a gaulish legend on a coin, written in ancient greek script?

If you want to combine "regular" registered subtags in the ways defined in
RFC 4646 they all need to be registered.  The "cel" is okay.  The "GRK" is
a typo, the registered scripts have four letters like "Grek".

And cel-Grek-gaulish would be "well-formed" for the reasons you've stated.

But gaulish is no registered variant at the moment, therefore cel-Grek-gaulish
is "invalid".  The grandfathered tags like cel-gaulish can't be combined with
other subtags.

If for some reason a variant gaulish is registered later the old "cel-gaulish"
would be reclassified as "redundant" instead of "grandfathered", and then the
tag cel-Grek-gaulish would be "valid" (= consisting of registered subtags in
the proper positions).  How to find any cel-Grek-gaulish document is another
question, not all "valid" tags make sense... :-)

Back to John's list, the irregular grandfathered tags will stay grandfathered
forever, there's no way to register all their parts for a reclassification as
redundant.  A simple irregular case (of the 17) is en-GB-oed, the "oed" part
is too short for a hypothetical variant subtag, unlike "gaulish".

Frank

Received on Wednesday, 8 November 2006 15:43:31 UTC