Behdad Esfahbod
Behdad Esfahbod
FontTools counterpart: https://github.com/fonttools/fonttools/pull/3395
This renders things now! Need to clean up a bit: remove the old glyf1 varcomposites, add test, benchmark, etc.
> This renders things now! > > Need to clean up a bit: remove the old glyf1 varcomposites, add test, benchmark, etc. glyf1 varcomposites removed. Just needs a test and...
Main TODO is subsetting closure over VARC, and subsetting VARC itself... Left for future.
Now with cycle detection, depth limit, and edge limit.
Both this and the fontttools PR are now up to date with the latest changes I made to the VARC table (including conditional components); and are have tests too. I...
> This is way above my head, so I have nothing meaningful to add. Thanks for checking. It should be straightforward translation of the spec: https://github.com/harfbuzz/boring-expansion-spec/blob/main/VARC.md
So.. I make mark attachment immediate in both directions? I think that's what the other engines do. Lemme put a PR together. As I said, that should make crowbar work...
The interaction with mark-advance-zeroing is annoying. I'd need to know that to decide whether to subtract the advances of the intervening marks when attaching a mark to a base.
This is my current working patch that fails only four tests: ```diff diff --git a/src/OT/Layout/GPOS/MarkArray.hh b/src/OT/Layout/GPOS/MarkArray.hh index 0887cc158..888f3f9c5 100644 --- a/src/OT/Layout/GPOS/MarkArray.hh +++ b/src/OT/Layout/GPOS/MarkArray.hh @@ -46,12 +46,15 @@ struct MarkArray :...