Re: Unicode 3.0 layout suggestion

From: Kenneth Whistler (
Date: Wed Aug 25 1999 - 15:25:58 EDT

Markus suggested:

> Suggestion to the production team of Unicode 3.0 and ISO 10646:2000:
> It would be *really* useful, if the example glyphs were shown together
> with a couple of reference lines, such that the alignment of the example
> glyphs can be recignized more easily.
> These reference lines should (at least for the European scripts and the
> symbols) be:
> - base line
> - height of x
> - height of H
> - left margin
> - right margin
> The reference lines could be made grey and very thin, such that they do
> not disturb the visual appearance of the glyph. Alternatively to the
> reference lines, you could also place short marker strokes on the glyph
> cell rim where the reference lines would end, such that I can use a
> ruler to add the line myself with a pencil.

Well, Michael Everson already responded correctly on this. But to forestall
Asmus springing a gear loose, I'll verify that there is a snowball's chance
in ... well, let's just say it isn't going to happen.

1. The layout for both Unicode 3.0 and ISO 10646:2000 charts is done
with custom-designed software that already has all kinds of little tweaks
in it to make oddball glyphs fit in their boxes without crashing. The
point is to get representative glyphs for the characters visible, without
running over the cell bounding boxes -- not easy, given the mixed collection
of scripts, and the disparate font sources we've had to gather to get
this thing printed. This code cannot and will not be tweaked further
between now and publication for *any* new feature additions; it is
just too risky--for both standards.

2. The Unicode Standard is a *character* standard, and not a font or
glyph design standard. We recognize the fact that people are going to
turn to it anyway and treat it as a font design benchmark--so we (and
Asmus in particular) have gone to great lengths to try to improve the
fonts and font design over Unicode 2.0. But we are putting an even more
prominent disclaimer in the Code Charts for Unicode 3.0: "Characters
images shown in the code charts are not prescriptive. In actual fonts
considerable variations are to be expected."

3. The information regarding any normative baseline, bounding box,
x-height, cap-height (or any other relevant font metric for other
scripts) is not a-priori available to us. Yes, usually this information
can be extracted from the fonts we use to print, and *could* be displayed
with a glyph (although there would be lots of complications for
combining characters, since we artificially construct their display
in the code charts), but that would simply be regurgitating what
a particular font designer's decisions were for the font we happen
to be using. Trying to analyze all that and determine what these
values *should* be for each character in the standard goes way beyond
our charter *and* our competencies.

> Unicode contains quite a number of characters, where font designers have
> no other easily accessible sources for learning how the intended
> alignment of the characters should be. For instance, which of the
> geometric shapes are supposed to sit on the baseline and which are
> supposed to fill the entire character cell without any safety distance
> from neighbor characters? Visible baselines would reduce the amount of
> guesswork and random decisions that the font designer has to make here
> and the resulting fonts will be more exchangeable. All to the benefit of
> the user, without restricting the font designers' freedom to chose
> whatever alignment seems artistically apropriate, if they are knowing
> what they are doing.

Please feel free to organize the Consortium of Unicode Font Designers,
which can fund and debate this effort, and establish glyph and font
design standards to assist in font interchange and compatibility.

I'm not being *completely* facetious here -- this is actually a
major undertaking, and I believe the expertise really lies mostly among
the professional font designers. The Unicode Consortium can help in
identifying what the oddball characters are supposed to be, but cannot
really specify in detail how they should look in a font.


> Markus
> --
> Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
> Email: mkuhn at, WWW: <>

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:51 EDT