Re: character by name on the gnome canvas, custom fontencoding

[Werner LEMBERG]

> Well, any time you want to run the Pango layout process, you need
> Unicode code points. So there needs to be a PUA mapping.  If the
> application author and the font creator can agree on a PUA mapping,
> that's the ideal situation.

Given that TeX users often have non-standard fonts/encodings, a clash
is much more likely than with standard PS fonts.

> >   I suggest
> > functions like
> > 
> >   glyphname_pua_mapping = pango_add_custom_encoding (<afmfile>);
> 
> My interest in adding questionable concepts like AFM files to the
> Pango API is precisely zero.

:-) BTW, why do you think that AFM files are bad?  Or is your main
concern that those files are separate from the glyph data?

> If fontconfig takes responsibility for locating the right AFM file,
> I'm willing to put in a little internal glue to call FT_Attach_File,
> but that's really all.

As I mentioned in my other mail, the glyph name info is in the PS font
also, so you can forget the AFM file for that (but not for composite
glyphs and ligature information).

> As my earlier mail indicated, there will be some facilities to allow
> apps to achieve this in the next version of Pango (though
> considerably more flexible, lowlevel, and correspondingly harder to
> use.)

Very good.

> But I was trying to suggest ways that current Pango versions could
> be used, and that basically means encoding the PUA mapping in the
> font in some fashion.  With a OpenType/CFF font it should be
> possible to have a Unicode mapping table and also maintain
> descriptive glyph names.

As a temporary solution this is definitely doable; fontforge's
scripting capabilities should make it straightforward to construct
non-overlapping PUA values for all lilypond fonts.


    Werner