we already have a header Language: defined in an IETF standards-track
document - RFC 1327.
Clearly I should have picked this up. Sorry.
Having now gone and looked again at 1327, it looks as if we might
want to do a wee bit of harmonization here, e.g., possibly moving
toward "Content-language" rather than "Language".
But the fact that this is needed in X.400 mapping strengthens, I think,
the argument for going toward language tagging per body part and in a
separate header, rather than, e.g., overloading it onto the character
set coding name.
NOTE: As I learned to my pain, linguists think that ISO 639 define
approximately 10 % of the world's languages, and a lot of them wrongly.
Hmm. I'm using 639 in a data interchange application and would have
estimated, based on the feedback we received while working on it and
since, more like 80% and only a few wrongly. But we were looking
strictly at languages in active use and, indeed, at languages that were
either the official or dominant language of some country.
We did provide an escape mechanism for describing languages not
listed, but it hasn't been used. Yet.
If one starts looking at historical languages, languages used only
in small enclaves, and languages that lack a written form that is
actually used by native speakers, I wouldn't be surprised to find that
things deteriorate in a hurry.
But, in practice, the languages needed for network communications may
be about the same subset of "all the world's languages" as my data
interchange application: While I haven't done a search since 10646 DIS1,
I suspect that 639 would cover all of the languages whose characters are
reflected in 10646 reasonably well. And the unwritten ones probably
aren't going to show up in email very soon :-)
it seems that linguists are as quarrelsome as networkers!
While they are a lot more polite, a TC46 meeting can be a really
revealing experience. That community seems, when it gets going, to find
far more nits to pick, and to pick them in much greater depth. :-)
--john