Yes and No. IMHO, it depends. I think that there's enough wars
around this problem already.
We think that supporting unification is a Good Thing, as well as
supporting multiple character sets is Good Thing. In fact, we've
implemented unificaiton table in forthcoming Emacs-20, so that one
could unify character in ISO 8859-1 and ISO 8859-3, or even to the one
in JISX208-1997.
I don't speak for specific implementations. I know that there are
many questionable points in exsiting applications/standards which
supports multiple character sets and/or unification. But I'd love
constructive way, and would like to make the points and goal clear.
My point is: don't implement the policy in lower layer, but do support
both, multiple character sets and unification, so that higher layer
(user) can choose one or both. (if possible)
If there are enough rationale for both sides, neglecting one side is
cause of wars. I don't like wars.
Thanks,
-- NIIBE Yutaka