@darius @migurski My assumption is that for something happening frequently from a long-lived process that an in-memory transform is going to be less resource intensive than forking. But I'm very, very out of practice here.