> Kenneth Whistler wrote on 2000-04-03 17:17 UTC:
> > Markus wrote:
> > > The question of what to do with other planes came up and there are
> > > various options, including but not limited to
> > >
> > > - finally doing the urgently needed fundamental revision of the X11
> > > architecture, which should provide for
> > >
> > > o 31-bit character set
> > You may want to also consider the *21*-bit option now.
> There are definitely places where using the 21-bit address INTERNALLY has
> real value.
> In the Ultracode bar code symbology standard, which is based on Unicode
> 3.0, input output data is specified as UTF-8 (default; since most bar code
> devices use an 8-bit data path), UTF-16 or UCS-4 to allow all characters
> on Code Planes 0-17 to be addressed. Internally we convert each UTF or UCS
> value to a "21-bit ISO address" for each character. Then we encode the
> address of the next character into the bar code based on the prior
> character address, in the minimum number of Mod 47 codewords, using a
> highly efficient encoding engine.
> This is especially efficient for encoding alphabetic languages on any Code
> Plane, where the actual character set used for bar code data tends to be
> <<47x47 characters, and often <47 consecutive characters (such as English
> or Cyrillic capital letters). In English, for example, 80% of the time the
> next character address is encoded one Mod 47 codeword. The key, however,
> is to work off the internal (difference of the 21-bit ISO addresses) mod
> Anyone wanting details on how and why we do it this way... E-mail me.
> Clive Hohberger
> AIM Technical Symbology Committee
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:00 EDT