[HarfBuzz] x_offset too small for a combining character

[Ondřej Majerech]

On 6 December 2016 at 05:39, Simon Cozens <simon at simon-cozens.org> wrote:
>
> On 05/12/2016 13:56, Ondřej Majerech wrote:
> > So the question is, what am I doing wrong in my program that makes
> > HarfBuzz report such low x_offset values?
>
> There's a couple of things here.
>
> The first is:
>
> >     std::cout << name_buffer << ": "
> >               << "x_advance = " << (positions[i].x_advance / 64)
> >               << ", y_advance = " << (positions[i].y_advance / 64)
>
> you're using integer math here. 64.0 will give you more precise answers:
>
> q: x_advance = 10.1562, y_advance = 0, x_offset = 0, y_offset = 0
> dotbelowcomb: x_advance = 0, y_advance = 0, x_offset = -1.09375,
> y_offset = -3.34375
> uni0307: x_advance = 0, y_advance = 0, x_offset = -1.28125, y_offset = 0
>
> But you're also confusing pixels and points. If you're really working at
> a resolution of 96 pixels per inch, as you say here:
>
> >   FT_Set_Char_Size(face, 12 << 6, 12 << 6, 96, 96);
>
> Then your dotbelowcomb needs to be adjusted by -1.09375 * 96 = -105 pixels.


So x_advance and y_advance are given in pixels, whereas x_offset and
y_offset are in points? Because I didn't multiply the advance values
with my DPI either, and the result looks okay. And even so, the -105
px figure sounds wrong as well: The q character is only 10 pixels
wide, moving back 105 pixels would put the combining dot 95 pixels in
front of the q.