Additionally it could be expensive if the library had to go through every binary sent from Elixir to check if it’s a valid UTF-8 encoded string and then convert it to Python string (and what if some binary data looks like a valid UTF-8 string accidentally?).
