Suggestion: 3.4 IntoIter - motivate the use of a newtype struct

[khs26]

Thanks for the excellent tutorial.

Sadly, Rust has nothing like a yield statement (yet), so we're going to have to implement the logic ourselves. Also, there's actually 3 different kinds of iterator each collection should endeavour to implement:

• IntoIter - T
• IterMut - &mut T
• Iter - &T

We actually already have all the tools to implement IntoIter using List's interface: just call pop over and over. As such, we'll just implement IntoIter as a newtype wrapper around List

I think could do with a bit more ramp up into why you use a newtype wrapper. I'm not sure what the best way to approach it is, but maybe the following questions/comments could help:

• Mention that we could just use self.pop(), but that the other iterator types will need to keep track of where they got to in the list.
• What is a newtype struct?
• When do we use a newtype struct?
  • Maybe gloss over the traits on foreign types stuff...

[achieveordie]

I second this opinion as the reader might assume that they can directly implement the Iterator trait on List<T>:

impl<T> Iterator for List<T> {
    type Item = T;
    fn next(&mut self) -> Option<Self::Item> {
        self.pop()
    }
}

and still pass the tests. This approach becomes problematic only when they progress to the next chapter and realize that they need separate implementations for iter() and iter_mut() using the same ergonomics.