Language vs. locale

Hi all,

I have created an issue
https://www.w3.org/International/multilingualweb/lt/track/issues/10
about "language versus locale". Currently the requirements doc. talks about
BCP 47 identifiers or ISO-639 values to identify locale or language. I
propose that we say:
- For language, we use BCP 47 identifiers. That includes fields for
language, region, script and variants, e.g. de-latn-de-1996 (German in
latin script, in Germany with new spelling conventions from 1996 - this
example is too verbose, see
http://people.w3.org/rishida/utils/subtags/index.php?check=de-latn-DE-1996&submit=Check,
but shows what you can express).
- For locales, we base our identifier on UTC #35
http://unicode.org/reports/tr35/
and say: we will make the conversion of "hyphen" to "underscore" as
described at
http://unicode.org/reports/tr35/#BCP_47_Language_Tag_Conversion
and we don't expect implementations of ITS 2.0 to understand the "u"
extension http://tools.ietf.org/html/rfc6067

Thoughts?

Felix

-- 
Felix Sasaki
DFKI / W3C Fellow