Steven G. Johnson

Note also that Symbolics.jl uses https://github.com/YingboMa/Unityper.jl (via SymbolicUtils.jl), which is another vote of confidence in using this to express a finite type union.

This package calls Qhull directly: https://github.com/gridap/MiniQhull.jl Maybe it should be merged with or replace Qhull.jl at some point?

Now that Qhull 8.0.999 has been released on Yggdrasil (https://github.com/JuliaPackaging/Yggdrasil/pull/3657) with the new accessors API (https://github.com/qhull/qhull/pull/89), it should be much easier to call libqhull_r directly (as also discussed in https://github.com/gridap/MiniQhull.jl/issues/5)...

I think that the goal would be to expose at least as much functionality as the Python API. The `facetT` and `vertexT` structures are simple enough to mirror in Julia...

@blegat, the qhull `Makefile` links the accessors API, but the Yggdrasil script uses `cmake` … I didn't realize qhull had both. I'll have to submit a patch for their `CMakeLists.txt`...

> I have used "facet_tail", "vertex_tail", "num_good", "del_vertices", "input_dim" that did not have accessors. `facet_tail` and `vertex_tail` are defined to have `id == 0` and `next == NULL`, so you...

Providing access to all of the fields seems excessive, since many of them are internal qhull state, but the scipy API seems like a good guide.

By the way, it might also be nice to have a `setNumber(name, val::Real) = set_number(name, Float64[val])` method, for convenience in setting scalar variables. (For some reason, the API defines a...

See also #27 — it's possible to use a mix of `.geo` files and Julia code.

The Python `yield` statement is only valid inside a Python function — that is, your `py"..."` block needs to include a whole function `def foo(...): ...` and not just a...