[HarfBuzz] Unicode vs glyphs

[Khaled Hosny]

On Tue, Jun 14, 2011 at 05:24:24PM +0200, Eduardo Castiñeyra wrote:
> El 14/06/2011 16:52, Shriramana Sharma escribió:
> >On 14-06-2011 20:18, Eduardo Castiñeyra wrote:
> >>Well, I was assuming that every glyph could be represented by a unicode
> >>character. So it looked more practical to have a list of unicode chars
> >>than a list of glyph indices because the last ones are font dependant.
> >
> >Hi -- I wonder what script you are rendering in which every glyph
> >could be represented by Unicode codepoints! See for example in
> >Indian scripts, lots of conjoining forms and ligatures do NOT have
> >codepoints. So I *think* that a rendering engine which is trying
> >to be able to support all scripts (?) should not assume that all
> >glyphs have Unicode codepoints because that is just not true!
> 
> To my knowledge, it seems to be true at least in latin, korean,
> hiragana, kanji, arabic, and some others we supported until now.
> Now, we begun to deal with harfbuzz and ICU because we also need to
> support Devanagari and Thai. So if these two scripts have
> non-codepointed glyphs I will have a good reason to force the
> redering guys to change their engine.

All the mentioned scripts can have fonts with glyphs that are not
assigned Unicode code points, you were lucky to not encounter them until
now, but in the era of "smart fonts" it is becoming more and more
common practice especially in high quality fonts.

Regards,
 Khaled

-- 
 Khaled Hosny
 Egyptian
 Arab