I think @twardoch requested this before. Google Fonts can probably use it too. We want functionality that analyzes GSUB/GPOS, and return sets of characters that all need to be in the same slice of the font. This is useful for slicing fonts for CSS serving optimization tricks.
behdad
All 5 comments
cc @garretrieger
behdad
on 10 Sep 2018
Let's resurrect this and implement in HarfBuzz and possible fonttools.
behdad
on 2 Jun 2020
FWIW here's a unionfind implementation I made quickly back in 2015 when @twardoch requested this feature from me privately: https://gist.github.com/behdad/0d36c27b6a664726a5b9c6050fc27d42
behdad
on 2 Jun 2020
I want to note that if we have a mark glyph, then it will bundle up all bases that define an attachment for that mark. Just making sure we all on the same page about that.
behdad
on 2 Jun 2020
Related to this we probably also want to consider codepoints that can participate in unicode compositions and decompositions as noted in this harfbuzz issue: https://github.com/harfbuzz/harfbuzz/issues/2283
garretrieger
on 2 Jun 2020
Related issues
ivanov-v
路
11Comments
typefacts
路
4Comments
cjdunn
路
11Comments
RomulBY
路
4Comments
arrowtype
路
8Comments
Most helpful comment
FWIW here's a unionfind implementation I made quickly back in 2015 when @twardoch requested this feature from me privately: https://gist.github.com/behdad/0d36c27b6a664726a5b9c6050fc27d42