Because '\u0000' is used in a special way in the
implementation of single-byte charsets,
for most such charsets (e.g. ISO-8859-2),
cs.canEncode('\u0000') returns false,
although ASCII is almost always encoded using the trivial and obvious
truncating conversion.

(This is a part of the fix for 5065777)

###@###.### 2004-06-22