Comment by dopu

Is set basically syntactic sugar for deepcopying a struct, mutating the specified field, and then returning that deepcopy? Seems like it could be quite slow.