RE: AFII font for UCS:2000

From: Marco.Cimarosti@icl.com
Date: Mon Oct 18 1999 - 06:29:56 EDT


James Agenbroad said:
>Could one hope to see a list (or lists by script?) of the sequences
>of Unicode codes that could invoke these "orphan glyphs"?

I agree with those who replied that such a thing cannot be done because
there are too many dependencies on fonts, language, typographic quality,
etc.

Take the example of Devanagari (encoded in Unicode), and imagine two
different applications of it:

#1: A professional typesetting system, tailored for Sanskrit, and used by
the top publisher of India to prepare the huge scholarly edition of the
Vedas for the years 2000's.

#2: A 2x20-cells liquid-crystal display installed on a refrigerator, showing
messages in Hindi such as: <Defrost needed>, <Current temperature: 4
degrees>, <Power was missing for: 23.45 minutes>.

How could these two applications share the same set of glyphs?

Application #1 needs all the possible ligatures and contextual forms that
ever appeared in a Sanskrit book since the creation of the Himalayas, along
with fuzzy logic and/or a complicated higher level protocol to select which
ligatures to use. The system on which the application runs is a dedicated
million$ workstation with virtually no memory (and budget) limitations. The
character:glyph ratio would be much bigger than your 1:3.

Application #2 would probably get along with a few "dead consonants", 3
forms or letter "ra" and (as a luxury item) the "ksha" ligature. The system
on which the application runs is a small 8-bit embedded processor with a
total 32 Kbyte RAM (that is busy doing a lot of other things, like
monitoring temperatures and humidity). The character:glyph ratio would now
be much smaller than 1:3.

And similar extrema could be found for any other script...

However, your idea becomes interesting if its scope is limited to "minimal
readability" (i.e., fitting application #2 or just a little bit more). In
this case, all the fine 'typographical nitpicking' that is often discussed
on this list could be ignored altogether (as it makes sense only on higher
quality systems), and it would become possible to define the "minimal set of
glyphs" needed for basic rendering.

A good example of this is the Arabic Presentation Forms B block (codes
U+FExx), that lists all the glyphs needed by a naive (Arabic-language)
implementation of the regular Arabic block (codes U+06xx). The same thing
could be done for other scripts, including the Indic alphabets.

But, why asking Unicode to do such a thing? (And who would volunteer to fill
in the request form and go through the process? :-)
Rather, there is a nice huge "Private Use Area" (codes U+E000 to U+F8FF)
that would be perfect for the purpose.

What I would suggest, is a private agreement among people who are interested
in simple "text-mode" applications like the Hindi fridge above (or, say,
Unicode for DOS-without-MS-Windows, or Linux-without-X-Windows, or pocket
databanks, and so on) to define this list of glyphs, encode it in the
P.U.A., and define the relevant algorithms to use them for basic rendering.

Regards. Marco



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:53 EDT