Derby still has some untested, unused code relating to a non-standard implementation of a Nationa Char type. The current code can be removed.
I believe the interesting functionality associated with this is now provided by
DERBY-1478 (territory based collation) . If Derby ever implements a
National Char type it should do so differently than the existing code, collation should not be tied to the National Char type.
I believe a future National char type might have to maintain a separate type id for compatibility with jdbc interface, but actual implmentation should be
the same code as the char types. Collating of the the national char type should be supported in exactly same way as regular char types.
If anyone is really intested in the national char code, it's history will always be available in svn, and a consistent version is available by looking at 10.0, 10.1,
and 10.2 codelines. I would propose any removal of code only take place in trunk and not be backported to a released codeline.