I don't think this is right. The process will crash, and the Supervision strategy you are using will determine what happens from there. This is what the BEAM is all about. The thing with NIFs is that they can crash the entire VM if they error.