Perfect. We'll want a document-level default for LANG which can optionally
get overridden for runs of characters.
>For the word importer the lid passed to the character handler will
>always be the correct one, though it is possible that currently I might
>be passing the wrong one occassionally but if this is true then it is a
>bug that will be found and fixed. But as far as I know the lid specifies
>the language, e.g. U.S. English, U.K. English and Australian English all
>exist, even though the charset is obviously the same.
Cool. I'm glad to hear that they go all the way down to the locale. That's
exactly what we'd want.