Behdad Esfahbod
Behdad Esfahbod
Sound right to me. Thanks for tracking it down.
> I think we should also drop the whole empty HVAR table, No. An empty HVAR table like the one you showed still have semantic value, it says: "advance widths...
If anything, that comment "# TODO Return emptiness..." in fonttools should be removed.
I'm not sure we want to support that. I think what we do in fonttools, at least in varlib when building a font, is to build once without AdvMap, and...
> The geometric mean of the last result is ~1.16, meaning Harfrust is now only 16% slower than Harfbuzz. It's even better than that. See the comment: https://github.com/harfbuzz/harfrust/discussions/257#discussioncomment-14363176 In short,...
> single glyph will work, but not multiple. That surprises me, as the code does this: ``` glyphs = args.glyphs.split() if args.glyphs else None ``` Did you try with a...
Lib/fontTools/varLib/interpolatable.py
Linking is going to be ugly. I'd have to move the per-shaper decomposition logic to a header, which I don't like, but probably the easiest. @jfkthame any opinions about the...
I've been testing this with: ``` $ hb-subset NotoSansBengali-Regular.ttf -u 995,9cb --output-file out.ttf $ hb-view out.ttf -u 995,9cb ```