[HarfBuzz] harfbuzz-ng: Branch 'master' - 2 commits

Khaled Hosny khaledhosny at eglug.org
Thu May 17 14:48:48 PDT 2012

On Thu, May 17, 2012 at 12:06:05AM -0400, Behdad Esfahbod wrote:
> On 05/16/2012 11:56 PM, Khaled Hosny wrote:
> > Without the intermediate glyphs the layout engine will not proceed to
> > the final glyphs.
> Ok, this is new to me.  Is it the case with Uniscribe?  I know HarfBuzz
> doesn't care, and from the code I've seen, Adobe doesn't care.

Really? I couldn't test this (no reliable subsetter), but I know realise
what I'm saying does not make any sense.

> > Obtaining initial and final glyphs is easy with hb-shape, what is
> > missing is the intermediate ones.
> Well actually initial glyphs is not *so* easy.  Are you thinking about setting
> a bogus script or what?
> >> But I see where you are coming from.  I can add this one too.  It's a bit of a
> >> different design from what I have right now, but I can probably add it by
> >> adding a generic /observer/ pattern to the GSUB/GPOS machinery, which kinda
> >> would be useful otherwise too...
> > 
> > The current tool fits fine for general subsetting, which is pretty
> > useful (it arrived just in time actually, I need it to make language
> > subset of my fonts :), but I'm talking about optimising special subsets.
> Yep.  I'm all for it.  Just need to understand the requirements and translate
> them to code :-).

After a bit of thinking, I think I don't really know what I'm talking
about (particularly how existing subsetters work), I'll trust you can
come with a proper solution for the use case I mentioned; subsetting a
font for very few words that is as smaller as possible while maintaining
working OpenType layout.


More information about the HarfBuzz mailing list