Around 15 o'clock on Jun 4, Markus Kuhn wrote:

> I get the impression that the MathML font system still uses a huge
> number of private use area characters to map its fonts to Unicode.

You're right, the PUA codepoints are then mapped to known locations in
known fonts; there's no intermediate standard representation for the 
glyphs.


> This was necessary before Unicode 3.2 was published, but most of the
> required characters are now in Unicode 3.2 and therefore, if you add
> embedded encoding tables for Xft, please map the fonts directly to Unicode
> 3.2 and let the Mozilla MathML people move to Unicode 3.2 as well, as
> opposed to implementing any temporary PUA mappings they might have come up
> with.

I don't particularily mind having Mozilla use PUAs internally as I believe
at least some of these glyphs are outside of the BMP.  Mozilla is creeping 
towards support for UCS-2 surrogates, but it's not there yet.

At the Xft interface, I'd like to see the PUAs converted to Unicode and 
then use the standard Xft Unicode interfaces -- either the 32-bit API or 
possibly a UCS-2 interface with surrogate support.

At the font level, I can use the Type1 tables to map names to Unicode
codepoints for fonts which don't yet support Unicode 3.2 mappings for the
new entries.  Fonts which don't provide either mapping may need extra
support, I'd rather do that in the form of an external application so that
the results can be checked manually, but I suspect we'll end up with some
hard-coded tables with knowledge about specific fonts inside fontconfig
(yuck).  I need to unify the Adobe glyph names with the MathML glyphs and 
the Unicode 3.2 spec.

Keith Packard        XFree86 Core Team        HP Cambridge Research Lab


_______________________________________________
Fonts mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/fonts

Reply via email to