Position paper for WWW i18n workshop in WWW5.

1. Charset
   Now, charset parameter is not added in ``Content-type:'' header.
   • Difficult to distinguish character code. Such as Latin-1 and EUC-JP, EUC-KR, EUC-CN, Shif-JIS and Big5.
   • No good charset for multilingual document.
   • Accept-charset become long and increase overhead.
2. Character code
   In Japan, many localized character code is used. These are not good for multilingual in one document.

• Change default character encoding.
  We propose to change default to ISO-2022 and Unicode.
• Sample implementation for multilingual document. http://www.wg.omron.co.jp/~shin/Arena-beta1d-CJK-doc(Arena i18n).

• RtoL, UtoD(vertical) rendering.
• More languages support, such as Thai, Vietnam. (these languages need ligature at rendering)
• Efficient folding mechanism for CJK.