Mark Crispin writes:
Can't we do an end run by stating that an implementation of the LANGUAGE extension MUST allow at least one language other than i-default?
Isn't that what Tim suggests? Arnt