"Use Unicode" is not easy solution for CJK

UTF-8 (UCS-2) is used in the software internal.
UTF-8 is *not* used "outside" (documents/printing/displaying/...)
We can't convert some characters irreversible:
EUC-JP -> Unicode -> ShiftJIS -> Unicode -> EUC-JP
It becomes some problems to use iconv: ex) samba
CJK Unicode convertion tables are a bit different between each vendors.