Cloning a hashmap using a reference to another hashmap

Question

I have a reference to a hashmap (data in the code below) which I want to clone into a new, owned hashmap. Cloning the reference gives me a new reference, which is not what I need.

I also tried doing an iter + map over the data reference and cloning the key and value pairs individually followed by a collect, but that also does not work. Here is a minimal working example:

use core::cell::Cell;
use std::collections::HashMap;
use std::collections::HashSet;
use std::rc::Rc;

struct Dummy<K, V> {
    dirty: Rc<Cell<bool>>,
    data: Cell<Option<HashMap<K, HashSet<V>>>>,
}

impl<K, V> Dummy<K, V> {
    fn persist(&self, prefix: &str, data: &HashMap<K, HashSet<V>>) {
        self.dirty.set(true);
        self.data.set(Some(data.clone()));
    }
}

which gives the following error:

error[E0308]: mismatched types
  --> src/lib.rs:14:28
   |
14 |         self.data.set(Some(data.clone()));
   |                            ^^^^^^^^^^^^ expected struct `std::collections::HashMap`, found reference
   |
   = note: expected type `std::collections::HashMap<K, std::collections::HashSet<V>>`
              found type `&std::collections::HashMap<K, std::collections::HashSet<V>>`

(Permalink to the playground)

The purpose of this code is to observe the contents of the hashmap on through the Dummy struct, which is used for unit testing.

I guess this problem is because given the generic types there is no way to determine how to deeply clone the key and value objects?

Is there a way to create a new hashmap given a reference to an existing hashmap?

What if you use impl<K: Clone, V: Clone> ?

b0fh
– b0fh

2019-12-17 12:04:41 +00:00
Commented Dec 17, 2019 at 12:04 — b0fh
– b0fh, Commented Dec 17, 2019 at 12:04

user2722968 · Accepted Answer · 2019-12-17 12:22:20Z

As the comment suggested this is a slightly unfortunate interaction of how Rust looks up methods via references / autoderef:

The implementation of Clone on HashMap requires the keys and the values (and the hasher) to be Clone as well. This makes sense since we can only clone the HashMap if we can clone its content. If any of those do not implement Clone, the HashMap does not implement Clone.

Your impl<K, V> has no bounds regarding Clone. So why does the call to data.clone() even work at all? That's because there is an implementation of Clone on any reference, which gives you a copy of the reference.

Since your impl<K, V> has no bounds on K or V, the compiler will only find the Clone-implementation on the primitive &-type to be applicable. If you require K: Clone, V: Clone, the HashMap's own Clone-impl applies (S is already Clone by default).

See this example:

// This returns a cloned HashMap
// because Rust uses the Clone-implementation on HashMap
fn do_clone<K: Clone, V: Clone>(data: &HashMap<K,V>) -> HashMap<K, V> {
    data.clone()
}

// There is no viable Clone-implementation on the HashMap
// because K & V are not Clone. But there is an implementation
// on the primitive reference type.
fn do_clone_ref<K, V>(data: &HashMap<K,V>) -> &HashMap<K, V> {
    data.clone()
}

Collectives™ on Stack Overflow

Cloning a hashmap using a reference to another hashmap

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related