For Java and other languages where identifiers can contain
multibyte/unicode characters we need to define how this should be
implemented.

With DWARF 1 we (DIAB) defined FORM_STRING to be a UTF-8 encoding of
the identifier. UTF-8 is specified by the Unicode consortium and has
the following properties:

1. UTF-8 strings are null terminated byte arrays.

2. UTF-8 strings use the most significant bit to indicate multibyte
characters.

3. UTF-8 strings that contain only 7 bit ASCII characters look exactly
like an ASCII string.

Adopted.  Description of "string" in Sections 2.2 and 7.5.4 should specify UTF-8.  Reference should be given to ISO/IEC 10646 AMI.