possible problem with JNI GetStringUTFChars

[Martin Buchholz]

Recommendations:
- open source the moribund jni spec, perhaps as part of openjdk, so that
people can improve it
- Add Scare doc to anything that deals with Modified UTF-8
- Add a Charset so that Java code can explicitly encode to Modified UTF-8,
despite being a bug magnet.
- AFAIK the "jnu" utility native functions are not public - they contain
useful stuff for efficiently converting between java char sequences and
native strings, e.g.
JNU_GetStringPlatformChars WITH_PLATFORM_STRING.  Creating public versions
of these might not be easy due to system differences, but it would be
worthwhile.