Re: Unicode repertoire of X11 fonts

From: Markus Kuhn (
Date: Mon Dec 07 1998 - 07:45:24 EST

Erik van der Poel wrote on 1998-12-07 06:09 UTC:
> The iso10646-1 fonts use a one-to-one character-to-glyph mapping, right?

Yes. Every BDF character with some number contains one a glyph
representing the Unicode character with the same code number. This seems
to be fully sufficient for the Latin/Greek/Hebrew/Armenian/Gregorian/
Cyrillic, etc. scripts, which do not have variant presentation forms. I
do not think that the X11 BDF mechanism is suitable for the more
complicated Indic and Arabic scripts. I think that major work has to be
done on the X11 font mechanism to support these properly, so this is
outside the scope of my -fixed-misc-* project. The same fome combining
characters. I am already happy if we can extend all the existing fixed
fonts to WGL4 or MES-2, I guess the more advanced stuff (e.g., a new
comprehensive Unicode output method as a server extension) will have to
be left to the Open Group and commercial supporters.

> Or do you somehow support true character to glyph mapping?

No. I wouldn't know how to do this, and the applications that mainly use
the fixed fonts (xterm, emacs, etc.) don't know how to do this either,
as there is no standard for a glyph encoding. So I have to use character
encoding = glyph encoding, which works fine for the European, Japanese,
Chinese, most African, and American scripts.

> > We left the option for future
> > standardization to indicate a subset (such as MES2) in the
> So, until the subsets are standardized in the future, the application
> programmer needs to look at the per_char array (if that info is
> required).

Yes. When I registered *-iso10646-1 in the XLFD scheme, I felt that it
was too early to standardize subsets as well, as ISO standards for
subsets are very early in the queue (especially the CEN/TC304/MES stuff
for European subsets). We should have another look at this topic in
summer 1999, when the ISO standards on MES will have gone through the

> Some applications need to know which glyphs are available. For example,
> in CSS you can say e.g.:
> font-family: helvetica, mincho, serif
> The implementation is supposed to go down this list until it finds a
> font that contains the glyph that you need. The generic names such as
> "serif" are themselves typically defined in terms of yet another list of
> fonts.

OK, makes sense, at least in the interim period in which many fonts will
have a much smaller repertoire than what people might want to have.


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 - 17:20:43 EDT