2 ms·

This is neat, but doing this for a whole module sounds really slow and prone to races if there’s threading