The classes sun.io.{ByteToChar CharToByte}Converter provide a mechanism for adding character encodings.
There is no other simple way of doing this. Keep the 
implementations of the standard character encodings part of
the implementation, but there should be a way to extend the 
character encodings by creating and registering an implementation
of the new encodings with a standard class.
(Review ID: 83223) 
======================================================================

Reimplement the character encoding lookup and registrations
in user code, wrap the builtin character encodings, and avoid 
all of the methods that take a builtin character encoding 
string.
======================================================================

Will be addressed under RFE 4287465.
  xxxxx@xxxxx   2000-01-26

Please add this capability. I have cases where
the encoding is unique for a given application and
doesn't agree with any of the standard encodings.

Fwiw, Bug 4228207 is a related but different 
request. There the builtin encoding classes
are more exposed. Here, it is possible to define
new encoders.