Closed cmyr closed 1 month ago
It turns out I have completely misunderstood this issue; the problem wasn't that fontc was treating these as bases but that it wasn't (and fonttools was; I misread the diff) and the root cause wasn't the absence of unicode values but the absence of explicit glyph categories; fix shortly.
I don't fully understand this issue yet, but fontc is treating some glyphs as bases where fontmake is not; all of the glyphs in question have the property that they are not assigned a unicode value.
Do we compile these glyphs? Are they reachable via substitution? Should we be filtering the input glyphs for mark generation at some earlier stage?