Re: Towards some more Private Use Area code points for ligatures

From: John Hudson (
Date: Fri May 31 2002 - 15:19:42 EDT

At 01:01 5/31/2002, William Overington wrote:

> >>I am also having a look at the idea of having code points for the famous
> >>combination border of the type used by Robert Granjon in the sixteenth
> >>century.

To which Michael Everson replied:

> >Code points are assigned to characters. Even in the PUA, it would be
> >advisable to keep to the character-glyph model.

To which William Overington responded:

>I am not sure that I understand what you mean. Does the following explain
>satisfactorily or am I missing some other aspect of the matter?

and then wrote 569 more words.

I believe what Michael is saying is that codepoints are normally assigned
to entities with semantic meaning, and that the encoding of something like
border units in an ornament system blurs the line between characters and
glyphs. This is a tricky area, and one that font developers face when
trying to figure out how to handle existing ornament or pi fonts in
Unicode-based formats like OpenType. Personally, I take the view that any
mapping of ornaments to characters, directly or indirectly (e.g. via
OpenType Layout lookups) is necessarily arbitrary, because there is no
logical relationship between these semantically empty glyphs and any
semantic character. Such fonts exist only because fonts are a convenient
storage and display mechanism for small monotone images, especially if such
images are intended to be composed in a manner similar to setting type or
to be used as in-line elements in text. But there is nothing fundamentally
typographic about them: they are an historical accident. Since the
relationship of ornament glyphs to character codepoints is arbitrary, the
only way to compare the benefits of different approaches is to consider
things like input models, user experience, etc.

The majority of existing (8-bit) ornament fonts use ASCII codes for
ornaments, often arranged in such a way that, in the case of border units,
there is a logical layout on (US) keyboard. For example, the keys Q W A S
might access four corner units, or rotated versions of the same element.
For a very large number of users, this is expected behaviour, so one
approach in OpenType and other Unicode-centric formats is to treat
ornaments as glyph variants of the same ASCII characters as in pre-existing
versions of the font. The benefit of this system is that it provides
immediate keyboard input support and a familiar user experience. Of course,
the 'text' set in this way can only be correctly displayed in the one font
with the appropriate layout feature turned on, but this is the very nature
of ornament fonts. William's proposed and, in my opinion, totally pointless
PUA encoding for Granjon ornaments has exactly the same limitation, but
lacks the immediate keyboard support and familiar user experience. The
wheel, once again, does not need reinvention, particularly not with corners.

John Hudson

Tiro Typeworks
Vancouver, BC

When the pages of books fall in fiery scraps
Onto smashed leaves and twisted metal,
The tree of good and evil is stripped bare.
                                        - Czeslaw Milosz

This archive was generated by hypermail 2.1.2 : Fri May 31 2002 - 13:41:06 EDT