Adding 'clone to satisfy the borrow checker' anti-pattern

[liamzdenek]

I feel like this write-up could be improved, but I'm not exactly sure how -- recommendations and modifications are welcome.

[cbreeden]

I like your motivation and description. I think we can make improve this section quite a bit if we put a little more polish into the example, and add examples that shows how you can avoid the problems. I'll play around with a few ideas and get back to you.

[cbreeden]

So I tried to think of some good examples, what do you think about this one

// Our personal key-value storage for names and ages
let mut vec = vec![("Bob".to_string(), 17), ("Stacy".to_string(), 21)];
let new_pair = ("Ferris".to_string(), 13);

// check for duplicates
let (ref key, _) = new_pair;
if vec.iter().map(|&(ref k, _)| k).find(|&k| k == key).is_none() {
    // Can't move since new_pair is being referenced still!
    vec.push(new_pair);
}

println!("{:?}", vec);

I could imagine someone trying to add a .clone() to the new_pair in vec.push(new_pair);. A lot of times these problems can be solved by properly managing their lifetimes -- in this case just adding k == &new_pair.0. In other cases, closing a mutable borrow scope in brackets. I don't know if this is the best example, just throwing it out there.

[llogiq]

I have a few ideas for examples:

• change an enum value, leaving parts intact (see #31)
• select & remove items from a collection (use Drain instead once available)
• un-by-ref an iterator (.cloned()) which is actually OK in many cases (unless you can consume the collection while iterating, which should be preferred)
• entry(_.clone()) – there's an RFC to get rid of that.

Hope that helps.

[pickfire]

Anything left for this?

[simonsan]

@liamzdenek any updates on this?

[simonsan]

Closing due to inactivity. Further additions to the PR can be made in #110.