[fa.laser-lovers] Large character sets and font file formats

LES@SU-AI@uw-beaver (LES@SU-AI) (12/05/83)

From: Les Earnest <LES@SU-AI>
The November 22 message from Wm. Leler, forwarded to Laser-Lovers on Dec. 1
by Brian Reid, inquires about large character set code standardization
and font file formats.  There are no widely accepted standards in either
area but such standards are badly needed.

Xerox is developing a character encoding standard as part of its Interpress
graphics language.  It uses existing international coding standards were
possible and includes many of the most widely used alphabets (Latin, Greek,
Cyrillic, Arabic, Hebrew, etc.), Japanese kana and kanji, and a number of
other widely used special symbols.  The coding scheme is constructed in
such a way that alphabetical material can be represented as strings of
8-bit characters, while ideographic languages such as Japanese would be
coded as 16-bit characters.

Imagen will be using a similar large character set based on 16 bit codes
that get mapped into smaller working code sets for any given printing job.
An internal memo describing this symbol set is available to anyone who is
interested.  (Yes, even people who buy Symbolics or Xerox printers).
Since network mail does not yet have any graphics standards, the only
practical way to send this material is via snail mail -- request from
LES @SU-AI or
  Les Earnest
  Imagen Corp.
  2660 Marine Way
  Mountain View, CA  94043
The working title for this character set is GASCII, for Galactic
Accumulative Standard Code for Information Interchange.  It is intended to
be a local standard covering only the Milky Way Galaxy.

On the font file format front, there is NO widely used format that can
practically work with large symbol sets.  Even formats that leave room for
(say) 16 bit character codes, require character directories that span the
code space.  Such directories are quite inefficient, especially for fonts
that use only a sparse subset of all possible codes.  We are developing a
new family of font file formats that use hash-coded directories to get
around this problem, but there is no documentation in a releasable form
yet.
        Les Earnest
        Imagen