Behdad Esfahbod comments

Results 1487 comments of


                                            Behdad Esfahbod

[aat] Insufficient test coverage for Indic AAT shaping

Note that the regression was fixed in 11.0.1.

CJK punctuation collapsing as part of shaping?

CC @jfkthame

CJK punctuation collapsing as part of shaping?

> Other technical challenges as far as I understand are: > > * The positioning engine needs to know the original Unicode code point, which the current HarfBuzz doesn't carry...

Avar2 support in font identification and contour variations / situations where coords are provided

Yeah, applying the ItemVariationStore only. It's this code: https://github.com/harfbuzz/harfbuzz/blob/f35b0a63b1c30923e91b612399c4387e64432b91/src/hb-ot-var-avar-table.hh#L318-L347

Optimise LangSys creation in feature compilation

The reason this optimization exists in the subsetter is that LangSys items that differ before subsetting might become identical after subsetting. A good font compiler would not create identical LangSys'es.

IUP optimisation should compare points and deltas as `int`s

I'm leaning towards `iup` code itself round when comparing for equality, hence keeping the error tolerance calculations to the original unrounded points. Only if the overhead is negligible...

Merge CFF with a CFF2 converted to CFF fails

Thanks for trying. The CFF converter is very brittle currently. Moreover, it produces a CID-keyed CFF, which I *think* is not supported by the merger. Or maybe it's the other...

Merge CFF with a CFF2 converted to CFF fails

Unfortunately I don't think I'll have the time and knowledge to fix this any time soon. You might want to see if the AFDKO tools can convert a CID-keyed CFF...

Re-implement callbacks in JS

Any perf numbers?

Re-implement callbacks in JS

That's quite impressive!