The current Java SJIS/Unicode mapping is based on the Unicode Consortium
table that is not consistent with the mapping used by Microsoft products,
such as Japanese Windows 95. This inconsistency creates huge inconvenience
for Japanese text processing because majority of Japanese text processing
still relies on the built-in assumptions of character encoding, such as
full/half-width representation.

Also the mapping inconsistency between Java and Microsoft products creates
inconsistent visual representation on Microsoft platforms.