[Proposal] UTF-16x (Apr.30 1999)
==================================================================

Title: UTF-16x (UTF-16 extension)

       UCS transformation format, 16 bit form
       and Upper compatible format of UTF-16

Introduction

  This document represents UTF-16x format.
  UTF-16x means UTF-16 extension,
  which can access all planes of ISO-10646-UCS-4.
  This format is constructed by 16bit quantities.

Background

  There are the following backgrounds to propose this format.

  * If The Unicode Standard aims to conform with ISO-10646,
    it needs to be able to process Plane 17 - 32867.
  * Many text data which are described by UCS-2/UTF-16 are
    already circulated, so it is impossible to ignore these data,
    and it is difficult for all data to be uniformed to UTF-8,
    when The Unicode Standard defines the character in Plane 17
    by any chance
  * Some research society already inspect to use ISO-10646 private
    use area ( Plane 224-255, Group 96-127 ).
    If they circulate these data which contains the code point over
    Plane 17 by UTF-8, the serious conversion problem will happen
    between UTF-8 and UTF-16.
  * If The Unicode Standard does not define conversion rule
    between ISO-10646-UCS-4 and UTF-16x,
    It is probable that vendors define nonstandard conversion rule
    each other, and The Unicode Standard becomes difficult to conform with
    ISO-10646 by this confusion in future.

Purpose

  * UTF-16x must have format of upper compatibility of UTF-16.
  * UTF-16x must be correctly converted form ISO-10646-UCS-4.
  * UTF-16x does not aim to be implemented by all processors.
  * UTF-16x need not be rendered by processors.
  * UTF-16x does not aim to be processed efficiently
  * UTF-16x may be used to distinguish ISO-10646 text
            from The Unicode Standard Text easily.

UTF-16x definition

  It applies same rule as UTF-16
  to the code points from 0x00000 to 0x10FFFF.

  But some code points are reserved for super surrogates.
  For example,
    super high surrogate   : 0x000EE000-0x000EE7FF
    super middle surrogate : 0x000EE800-0x000EEBFF
    super row surrogate    : 0x000EEC00-0x000EEFFF

  The code points from 0x00110000 to 0x7FFFFFFF are represented
  by super surrogate trio (high, middle, row).

  This means that UTF-16x sequence needs 12 octets and
  it certainly has the next sequence.
  (low surrogate : 0xDC00-0xDFFF)

  0xDB78 + low surrogate + 0xDB7A + low surrogate + 0xDB7B + low surrogate
    or
  0xDB79 + low surrogate + 0xDB7A + low surrogate + 0xDB7B + low surrogate

Encoding UTF-16x

  At first, binary expression is used.

  UCS-4(binary expression):
    0wxxxxxx-xxxxyyyy-yyyyyyzz-zzzzzzzz

  UTF-16x(binary expression):
    11011011-0111100w, 110111xx-xxxxxxxx,
    11011011-01111010, 110111yy-yyyyyyyy,
    11011011-01111011, 110111zz-zzzzzzzz

  Next, hexadecimal expression is used.

  UCS-4 range:
    0x00110000-0x3FFFFFFF

  UTF-16x expression:
    U+DB78 + low surrogate + U+DB7A + low surrogate + U+DB7B + low surrogate

  UCS-4 range:
    0x40000000-0x7FFFFFFF

  UTF-16x expression:
    U+DB79 + low surrogate + U+DB7A + low surrogate + U+DB7B + low surrogate

==================================================================
Masahiko Maedera, ???????????????????????
Masahiko_Maedera@???????????