Concrete subtags

Addison Phillips [wM] aphillips at
Thu May 29 14:04:35 CEST 2003

> John Clews on 05/29/2003 02:59:04 PM:
> > I quite like that idea of Addison's.
> You do? My reaction was that it sounded like it would mean we'd have to
> register *everything* short of ISO 639 IDs -- quite the opposite of having
> generative mechanisms as we have been talking about.
That's not what I meant. I meant that, in the event you want to register
something *other than* one of the standard tags in the generative mechanism
(in the current thinking that would be ISO639, ISO3166, and ISO15924), that
you would register it as be an amendment to one of these. That is, "this
subtag 'en-scouse' is a language tag" or "this subtag 'hans' is a script
tag" or "this subtag '1901' is an orthographic variation". Informative
information about the tag would explain that (for example) "1901" is
intended for use with the various generated German things (things with a
base of 'de'). But the distinctions could also be made on another level
where necessary (tag 'xxxxx' is an othographic variation that applies to

Then you can build tables that do useful things when both generating and
parsing. This is contrast to "de-AT-1901" (for example), which has to be
considered an atomic exceptional tag and then you have to know that you're
allowed (in this case) to "see inside" the tag.

Yes, I'm thinking like a programmer here, which, if experience holds true,
is probably not a good way to handle language tags ;-).


More information about the Ietf-languages mailing list