Re: [ft-devel] Hinting paths

[Nikolaus Waxweiler]

> One peculiarity to note: If in `native ClearType' mode, a modification
> of a glyph's advance width is properly executed by the MS hinter.
> However, the MS rasterizer ignores it, always using linear advance
> widths in modern Windows versions.  Previous versions supported a
> `compatible-width ClearType' mode (cf. selector bit 7 in the GETINFO
> instruction).  Unfortunately, there is no way currently to circumvent
> this.

So that's where the different advances between GDI and DirectWrite come 
from? :)

> Yes.  Backwards-compatibility mode is essential to hint older fonts,
> or fonts that are written to use the same bytecode for both ClearType
> and non-ClearType, relying on the fact that (most) instructions along
> the horizontal axis are ignored.

And that mode is hidden behind the SUBPIXEL_HINTING define? What's the 
difficulty in leaving out horizontal instructions?

> Example, please, ideally translated to key sequences that I can use
> with ftview.

Hm, I couldn't reproduce the differences between autohint off medium and 
full (reset my conf.d to mostly defaults yesterday, might have been some 
spurious interaction with 10-autohint.conf or something?). However, the 
difference between autohint on medium and full is still there: 
http://postimg.org/gallery/27mhq5d00 (full-noauto, full-auto, 
medium-auto). Not sure about the key strokes, according to cairo source, 
both medium and full should lead to FT_LOAD_TARGET_LCD(_V) and autohint 
should result in an additional FT_LOAD_FORCE_AUTOHINT. Dunno.