Proposal: Language code "de-DE-trad"

Harald Tveit Alvestrand
Wed, 13 Feb 2002 16:53:27 +0100

--On 11. februar 2002 09:31 -0600 wrote:

> These four registration requests look to me to be well-motivated, and the
> proposed codes are reasonable, with one possible concern: de-DE and de-AT
> are already in use designating data in the tradition orthographies. Does
> it make sense to use the shorter tags to mean the traditional
> orthographies, or should the shorter tags become understood to be
> ambiguous with regard to the orthographies?

we should differentiate between two different formalisms here, more or less 
similar to the difference between "accepting" and "emitting" grammars - 
with the latter usually being quite a bit stricter than the former.

I think the principle of least astonishment indicates that:

- when asking for data using a short tag in an "accept-language" header, 
any variant is acceptable, but if multiple ortographies are available, the 
latest one should be preferred (its q value increased over the others).
- when tagging data, if you care about ortograhpy, use long tags.