[HarfBuzz] harfbuzz.old applies x advances for rtl in the wrong direction

[Khaled Hosny]

On Thu, Jan 13, 2011 at 10:50:16PM +0100, Jjgod Jiang wrote:
> On Thu, Jan 13, 2011 at 8:45 PM, Khaled Hosny <khaledhosny at eglug.org> wrote:
> > "X and Y values specified in OpenType fonts for placement operations are
> > always within the typical Cartesian coordinate system (origin at the
> > baseline of the left side), regardless of the writing direction."
> 
> Note that just after this, "However, it's important to note that the
> meaning of “advance width” changes, depending on the writing direction."
> Since advance width depends on the writing direction, adjustment to
> advance width should depend on the writing direction as well, I assume.
> 
> Anyway, I hope someone more familiar with the spec can clarify this.
> 
> > I can't comment on this, but I just tested with FireFox 4 and got
> > correct kerning.
> 
> Firefox 4 hasn't use harfbuzz-ng for Arabic shaping yet.

It does since beta 8.

> > The issue is not limited to this font, I can reproduce it with other
> > fonts like Arabic Typesetting from MS and many Linotype fonts. Arabic
> > fonts from Apple are AAT fonts, so they don't apply here (may be
> > TextEdit on your system is not doing OpenType at all; Apple have been
> > too late to the game).
> 
> I may have used the AAT font, but TextEdit does support OpenType and
> Arabic shaping (in fact, Cocoa Text System supports a lot of OpenType
> features better than software from Adobe and Microsoft.)

I'm not sure about Apple stuff, but as far as I'm concerned the font is
OK and have been tested with 4 independent OpenType implementations
(Uniscribe, Pango/HarfBuzz, ICU and LuaTeX).

> I just verified it with Arabic Typesetting font, it works well here.

So with Qt you see no gaps between connected letters of the test string
I posted?

Regards,
 Khaled

> 
> - Jiang

-- 
 Khaled Hosny
 Arabic localiser and member of Arabeyes.org team
 Free font developer