(no subject)

[Jon Hanna]

> On the other hand:
>
> * All other things being equal (which they never are), if we did
> introduce
> orthography as yet another header, we could wait until the Sun goes red
> giant before any of the browsers/clients and servers actually implement
> content negotiation along this extra dimension,

Conversely however, the separating the ability to negotiate on the grounds
of language from the ability to negotiate on the grounds of
language+orthography is likely to be more easily compatible with older
systems. In particular recognising Old Irish in Ogham as a form of Old Irish
would require guesswork on the part of a system built currently (guesswork
based on treating language tags as hierarchical, which previous threads have
shown to be ill-advised).

> * People's orthographic/script preferences are almost always attached to
> particular languages (at least for languages they actually understand,
> which is 99% of what we should be interested in here).

That's a user-interface issue, not a code issue.

I would certainly like a simple link between language and script (using
natural language names rather than codes when possible) as a user. I would
much prefer nicely separated abstracts as a programmer. Going from one to
the other is what hardly a novel task in UI design.