Re: [ft-devel] Clean-up of af_loader_load_glyph()

[Nikolaus Waxweiler]

> > > > Aren't these conceptually incompatible with the autohinter method
> > > > of snapping outer edges to the grid to achieve sharpness?
> > >
> > > I don't think so, cf. the `warping' code in the auto-hinter.
> >
> > So this support would only be relevant for warping?
>
> No.  It's not related.  I just want to point you to the warper that
> does try to micro-optimize the horizontal positioning of a glyph.

Okay, but my question is, would this "subpixel hinting support" in the 
AH be relevant to LIGHT (Y-axis only) or NORMAL (X+Y-axis) mode? My 
guess would be no. LIGHT doesn't touch the X-axis, NORMAL by nature 
adjusts to pixel integers. So the warper would be the only use case? Or 
am I misunderstanding?

By the way, I found that the warper modifies advance widths. I guess you 
could compensate for that by twiddling the scaler deltas?

> > Come to think of it, hinting glyphs (native or autohinter) always
> > rounds advance widths to integer pixels iirc.
>
> Yes.
>
> > Wouldn't that need to change, too, for fractional advance widths?
>
> Correct.  It would be possible to adapt and re-use the warper code for
> non-integer positioning.  Whether this is useful, beautiful, fast
> enough, etc., etc., is another issue.

Well, wouldn't it be easier to stop rounding to pixels integers instead 
if the hinting doesn't change the advance widths anyway?