RE: Single Unicode Font

From: Ayers, Mike (
Date: Tue May 22 2001 - 14:03:24 EDT

> From: Carl W. Brown []

> I find that the most compelling reason is that many
> characters should be
> rendered differently depending on user preferences. For
> example a Japanese
> user should have the han rendered into Japanese characters
> except for one
> that do not exist in Japanese. These can be rendered with a
> Chinese font.
> Likewise do you use Simplified or Traditional Chinese glyphs?

        For those who do not know enough to tell the difference between
Kanji typography and Hanzi typography (and Hanja typography ;-) this yields
no benefit and forces a meaningless choice (which script that you can't read
do you prefer?).

> As you said font groupings are the best way to go for a
> "complete" font.
> With a series for fonts grouped together so that if the
> characters is not on
> the first fonts it selects from the second fonts, etc. is the
> way to go.

        I do not question the benefiot of such a system for getting the best
viewing experience. Having recently (and still, really, as it takes quite
some time) added multilingual support to emacs on Win2000, I found that even
with that OS's excellent multilingual capabilities, this is a painful thing
to do. For instance, there is no trivial way to determine what scripts are
supported by any given font - therefore I must keep modifying my fontset
specification, reloading it, and looking at the "hello" page again. All
that, and I am still limited to the scripts on the "hello" page (if I ever
get all them!).

> With Unicode extended planes and potentially millions of
> characters a single
> font make no sense. Fonts like Microsoft's Unicode fonts
> which handle most
> of the major scripts are potentially useful as a starting
> point. Beyond
> that I think that we all agree that it is insane to try to
> produce a full
> Unicode font.

        "Insane"? No, I was thinking words like "logical", "useful", and
"sensible", but I think that there is more agreement here than would at
first appear. I agree that an all-encompassing Unicode font would probably
have hideous aesthetic properties, but that does not detract from its
usefulness in rendering characters that I had not expected to receive. Once
I know what those characters are, I can then load a better quality font for
that script atop the all-purpose font and see the characters "as the sender
meant them to be seen". :-)

        Also, a note on the oft-cited issue of the size problems of an
all-purpose Unicode font. The largest font on my system is 16MB. Since it
is probably a CJK font, I amguessing that it covers 1/3 to 1/2 the Unicode
range (1/2 of 3.0, 1/3 of 3.1?) - to be safe, let's expand it four times, to
64MB. This size could still tuck neatly into the swap file of (almost) any
personal system you could buy today. Given the expanding hardware
requirements of newer OS's, this doesn't seem to me to be much of an issue.


P.S. To be absolutely clear, I don't care if I need to load many related
fonts (UniFont1, UniFont2, etc.) into a fontset to get complete coverage -
it's just the complete coverage that matters to me.

This archive was generated by hypermail 2.1.2 : Fri Jul 06 2001 - 00:18:17 EDT