The .push() method on Vec<T> takes as its argument a &mut Vec<T>, and the addition of the new element is a side effect.

What is the reason it was decided against push to be purely functional such that its API would look this:

fn push(mut self, elem) -> Self

Would taking ownership of the Vec<T> not be ergonomic in some way? I do not believe there to be a substantial performance hit by moving it, given that Vec is just a fat pointer.