CharToByte convert getMaxBytesPerChar() returns bytes necessary to hold native bytes for one java unicode character. Some converter, like, ISO2022JP will comsumes up to 5 bytes (including escape sequence). But getMaxBytesPerChar() returns '2'. Unless it is private API, sun.io, we can control how caller allocate output bytes buffer. However, this will be a problem when we make API to public in future.

This bug is derived from 4252798 that deals with two problem. So, I split it.
koushi.takahashi@japan 1999-07-14