Colin Rofls

I agree a common glyph name type is something we should have for the compiler, but I'm not sure it should be in font-types? It isn't really anything to do...

From a cursory look I think that fonttools is doing the second proposal: going per-delta-set, and using the compressed version if it is smaller: https://github.com/fonttools/fonttools/blob/main/Lib/fontTools/ttLib/tables/TupleVariation.py#L576

for the record the PR that added the build script was #161; the problem was that `include_bytes!` is expanded early in compilation, and will fail if a path is missing,...

I'm not sure it will be trivial to deny dependencies in CI, but it should be easy enough to do in code review, and we're a small enough set of...

I think part of what is going on here is a misunderstanding of parts of that function? In particular there are lots of uses of `expect`, but here I'm using...

I have failed to make myself clear. The conditions that would trigger a crash here is *in no way a function of the input*. Our code basically looks like this:...

It is impossible because we only call serialize *if packing is successful*. We define a "successful packing" as one for which, for every offset between two tables, the distance (in...

agree, we should not be casting.

This is definitely a minor annoyance; the names are taken from the names of the structs for each type, and that follows Rust's naming conventions. The top-level tables do now...

hello! Sorry for the sparse docs, I'll definitely get to that eventually, but things are still in quite a bit of flux. The answer provided in that question looks right...