Files
docs-rust/ch20/ch20-02-advanced-traits.html
2026-06-22 21:27:36 +05:30

711 lines
38 KiB
HTML
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Advanced Traits</title>
</head>
<body>
<h2 id="advanced-traits"><a class="header" href="#advanced-traits">Advanced Traits</a></h2>
<p>We first covered traits in the <a href="../ch10/ch10-02-traits.html">“Defining Shared Behavior with
Traits”</a><!-- ignore --> section in Chapter 10, but we didnt discuss
the more advanced details. Now that you know more about Rust, we can get into
the nitty-gritty.</p>
<!-- Old headings. Do not remove or links may break. -->
<p><a id="specifying-placeholder-types-in-trait-definitions-with-associated-types"></a>
<a id="associated-types"></a></p>
<h3 id="defining-traits-with-associated-types"><a class="header" href="#defining-traits-with-associated-types">Defining Traits with Associated Types</a></h3>
<p><em>Associated types</em> connect a type placeholder with a trait such that the trait
method definitions can use these placeholder types in their signatures. The
implementor of a trait will specify the concrete type to be used instead of the
placeholder type for the particular implementation. That way, we can define a
trait that uses some types without needing to know exactly what those types are
until the trait is implemented.</p>
<p>Weve described most of the advanced features in this chapter as being rarely
needed. Associated types are somewhere in the middle: Theyre used more rarely
than features explained in the rest of the book but more commonly than many of
the other features discussed in this chapter.</p>
<p>One example of a trait with an associated type is the <code>Iterator</code> trait that the
standard library provides. The associated type is named <code>Item</code> and stands in
for the type of the values the type implementing the <code>Iterator</code> trait is
iterating over. The definition of the <code>Iterator</code> trait is as shown in Listing
20-13.</p>
<figure class="listing" id="listing-20-13">
<pre><code class="language-rust noplayground">pub trait Iterator {
type Item;
fn next(&amp;mut self) -&gt; Option&lt;Self::Item&gt;;
}</code></pre>
<figcaption><a href="#listing-20-13">Listing 20-13</a>: The definition of the <code>Iterator</code> trait that has an associated type <code>Item</code></figcaption>
</figure>
<p>The type <code>Item</code> is a placeholder, and the <code>next</code> methods definition shows that
it will return values of type <code>Option&lt;Self::Item&gt;</code>. Implementors of the
<code>Iterator</code> trait will specify the concrete type for <code>Item</code>, and the <code>next</code>
method will return an <code>Option</code> containing a value of that concrete type.</p>
<p>Associated types might seem like a similar concept to generics, in that the
latter allow us to define a function without specifying what types it can
handle. To examine the difference between the two concepts, well look at an
implementation of the <code>Iterator</code> trait on a type named <code>Counter</code> that specifies
the <code>Item</code> type is <code>u32</code>:</p>
<figure class="listing">
<span class="file-name">Filename: src/lib.rs</span>
<pre><code class="language-rust ignore"><span class="boring">struct Counter {
</span><span class="boring"> count: u32,
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Counter {
</span><span class="boring"> fn new() -&gt; Counter {
</span><span class="boring"> Counter { count: 0 }
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>impl Iterator for Counter {
type Item = u32;
fn next(&amp;mut self) -&gt; Option&lt;Self::Item&gt; {
// --snip--
<span class="boring"> if self.count &lt; 5 {
</span><span class="boring"> self.count += 1;
</span><span class="boring"> Some(self.count)
</span><span class="boring"> } else {
</span><span class="boring"> None
</span><span class="boring"> }
</span><span class="boring"> }
</span><span class="boring">}</span></code></pre>
</figure>
<p>This syntax seems comparable to that of generics. So, why not just define the
<code>Iterator</code> trait with generics, as shown in Listing 20-14?</p>
<figure class="listing" id="listing-20-14">
<pre><code class="language-rust noplayground">pub trait Iterator&lt;T&gt; {
fn next(&amp;mut self) -&gt; Option&lt;T&gt;;
}</code></pre>
<figcaption><a href="#listing-20-14">Listing 20-14</a>: A hypothetical definition of the <code>Iterator</code> trait using generics</figcaption>
</figure>
<p>The difference is that when using generics, as in Listing 20-14, we must
annotate the types in each implementation; because we can also implement
<code>Iterator&lt;String&gt; for Counter</code> or any other type, we could have multiple
implementations of <code>Iterator</code> for <code>Counter</code>. In other words, when a trait has a
generic parameter, it can be implemented for a type multiple times, changing
the concrete types of the generic type parameters each time. When we use the
<code>next</code> method on <code>Counter</code>, we would have to provide type annotations to
indicate which implementation of <code>Iterator</code> we want to use.</p>
<p>With associated types, we dont need to annotate types, because we cant
implement a trait on a type multiple times. In Listing 20-13 with the
definition that uses associated types, we can choose what the type of <code>Item</code>
will be only once because there can be only one <code>impl Iterator for Counter</code>. We
dont have to specify that we want an iterator of <code>u32</code> values everywhere we
call <code>next</code> on <code>Counter</code>.</p>
<p>Associated types also become part of the traits contract: Implementors of the
trait must provide a type to stand in for the associated type placeholder.
Associated types often have a name that describes how the type will be used,
and documenting the associated type in the API documentation is a good practice.</p>
<!-- Old headings. Do not remove or links may break. -->
<p><a id="default-generic-type-parameters-and-operator-overloading"></a></p>
<h3 id="using-default-generic-parameters-and-operator-overloading"><a class="header" href="#using-default-generic-parameters-and-operator-overloading">Using Default Generic Parameters and Operator Overloading</a></h3>
<p>When we use generic type parameters, we can specify a default concrete type for
the generic type. This eliminates the need for implementors of the trait to
specify a concrete type if the default type works. You specify a default type
when declaring a generic type with the <code>&lt;PlaceholderType=ConcreteType&gt;</code> syntax.</p>
<p>A great example of a situation where this technique is useful is with <em>operator
overloading</em>, in which you customize the behavior of an operator (such as <code>+</code>)
in particular situations.</p>
<p>Rust doesnt allow you to create your own operators or overload arbitrary
operators. But you can overload the operations and corresponding traits listed
in <code>std::ops</code> by implementing the traits associated with the operator. For
example, in Listing 20-15, we overload the <code>+</code> operator to add two <code>Point</code>
instances together. We do this by implementing the <code>Add</code> trait on a <code>Point</code>
struct.</p>
<figure class="listing" id="listing-20-15">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024">use std::ops::Add;
#[derive(Debug, Copy, Clone, PartialEq)]
struct Point {
x: i32,
y: i32,
}
impl Add for Point {
type Output = Point;
fn add(self, other: Point) -&gt; Point {
Point {
x: self.x + other.x,
y: self.y + other.y,
}
}
}
fn main() {
assert_eq!(
Point { x: 1, y: 0 } + Point { x: 2, y: 3 },
Point { x: 3, y: 3 }
);
}</code></pre>
<figcaption><a href="#listing-20-15">Listing 20-15</a>: Implementing the <code>Add</code> trait to overload the <code>+</code> operator for <code>Point</code> instances</figcaption>
</figure>
<p>The <code>add</code> method adds the <code>x</code> values of two <code>Point</code> instances and the <code>y</code>
values of two <code>Point</code> instances to create a new <code>Point</code>. The <code>Add</code> trait has an
associated type named <code>Output</code> that determines the type returned from the <code>add</code>
method.</p>
<p>The default generic type in this code is within the <code>Add</code> trait. Here is its
definition:</p>
<pre class="playground"><code class="language-rust edition2024"><span class="boring">#![allow(unused)]
</span><span class="boring">fn main() {
</span>trait Add&lt;Rhs=Self&gt; {
type Output;
fn add(self, rhs: Rhs) -&gt; Self::Output;
}
<span class="boring">}</span></code></pre>
<p>This code should look generally familiar: a trait with one method and an
associated type. The new part is <code>Rhs=Self</code>: This syntax is called <em>default
type parameters</em>. The <code>Rhs</code> generic type parameter (short for “right-hand
side”) defines the type of the <code>rhs</code> parameter in the <code>add</code> method. If we dont
specify a concrete type for <code>Rhs</code> when we implement the <code>Add</code> trait, the type
of <code>Rhs</code> will default to <code>Self</code>, which will be the type were implementing
<code>Add</code> on.</p>
<p>When we implemented <code>Add</code> for <code>Point</code>, we used the default for <code>Rhs</code> because we
wanted to add two <code>Point</code> instances. Lets look at an example of implementing
the <code>Add</code> trait where we want to customize the <code>Rhs</code> type rather than using the
default.</p>
<p>We have two structs, <code>Millimeters</code> and <code>Meters</code>, holding values in different
units. This thin wrapping of an existing type in another struct is known as the
<em>newtype pattern</em>, which we describe in more detail in the <a href="ch20-02-advanced-traits.html#implementing-external-traits-with-the-newtype-pattern">“Implementing
External Traits with the Newtype Pattern”</a><!-- ignore --> section. We
want to add values in millimeters to values in meters and have the
implementation of <code>Add</code> do the conversion correctly. We can implement <code>Add</code> for
<code>Millimeters</code> with <code>Meters</code> as the <code>Rhs</code>, as shown in Listing 20-16.</p>
<figure class="listing" id="listing-20-16">
<span class="file-name">Filename: src/lib.rs</span>
<pre><code class="language-rust noplayground">use std::ops::Add;
struct Millimeters(u32);
struct Meters(u32);
impl Add&lt;Meters&gt; for Millimeters {
type Output = Millimeters;
fn add(self, other: Meters) -&gt; Millimeters {
Millimeters(self.0 + (other.0 * 1000))
}
}</code></pre>
<figcaption><a href="#listing-20-16">Listing 20-16</a>: Implementing the <code>Add</code> trait on <code>Millimeters</code> to add <code>Millimeters</code> and <code>Meters</code></figcaption>
</figure>
<p>To add <code>Millimeters</code> and <code>Meters</code>, we specify <code>impl Add&lt;Meters&gt;</code> to set the
value of the <code>Rhs</code> type parameter instead of using the default of <code>Self</code>.</p>
<p>Youll use default type parameters in two main ways:</p>
<ol>
<li>To extend a type without breaking existing code</li>
<li>To allow customization in specific cases most users wont need</li>
</ol>
<p>The standard librarys <code>Add</code> trait is an example of the second purpose:
Usually, youll add two like types, but the <code>Add</code> trait provides the ability to
customize beyond that. Using a default type parameter in the <code>Add</code> trait
definition means you dont have to specify the extra parameter most of the
time. In other words, a bit of implementation boilerplate isnt needed, making
it easier to use the trait.</p>
<p>The first purpose is similar to the second but in reverse: If you want to add a
type parameter to an existing trait, you can give it a default to allow
extension of the functionality of the trait without breaking the existing
implementation code.</p>
<!-- Old headings. Do not remove or links may break. -->
<p><a id="fully-qualified-syntax-for-disambiguation-calling-methods-with-the-same-name"></a>
<a id="disambiguating-between-methods-with-the-same-name"></a></p>
<h3 id="disambiguating-between-identically-named-methods"><a class="header" href="#disambiguating-between-identically-named-methods">Disambiguating Between Identically Named Methods</a></h3>
<p>Nothing in Rust prevents a trait from having a method with the same name as
another traits method, nor does Rust prevent you from implementing both traits
on one type. Its also possible to implement a method directly on the type with
the same name as methods from traits.</p>
<p>When calling methods with the same name, youll need to tell Rust which one you
want to use. Consider the code in Listing 20-17 where weve defined two traits,
<code>Pilot</code> and <code>Wizard</code>, that both have a method called <code>fly</code>. We then implement
both traits on a type <code>Human</code> that already has a method named <code>fly</code> implemented
on it. Each <code>fly</code> method does something different.</p>
<figure class="listing" id="listing-20-17">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024">trait Pilot {
fn fly(&amp;self);
}
trait Wizard {
fn fly(&amp;self);
}
struct Human;
impl Pilot for Human {
fn fly(&amp;self) {
println!("This is your captain speaking.");
}
}
impl Wizard for Human {
fn fly(&amp;self) {
println!("Up!");
}
}
impl Human {
fn fly(&amp;self) {
println!("*waving arms furiously*");
}
}
<span class="boring">
</span><span class="boring">fn main() {}</span></code></pre>
<figcaption><a href="#listing-20-17">Listing 20-17</a>: Two traits are defined to have a <code>fly</code> method and are implemented on the <code>Human</code> type, and a <code>fly</code> method is implemented on <code>Human</code> directly.</figcaption>
</figure>
<p>When we call <code>fly</code> on an instance of <code>Human</code>, the compiler defaults to calling
the method that is directly implemented on the type, as shown in Listing 20-18.</p>
<figure class="listing" id="listing-20-18">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024"><span class="boring">trait Pilot {
</span><span class="boring"> fn fly(&amp;self);
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">trait Wizard {
</span><span class="boring"> fn fly(&amp;self);
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">struct Human;
</span><span class="boring">
</span><span class="boring">impl Pilot for Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("This is your captain speaking.");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Wizard for Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("Up!");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("*waving arms furiously*");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>fn main() {
let person = Human;
person.fly();
}</code></pre>
<figcaption><a href="#listing-20-18">Listing 20-18</a>: Calling <code>fly</code> on an instance of <code>Human</code></figcaption>
</figure>
<p>Running this code will print <code>*waving arms furiously*</code>, showing that Rust
called the <code>fly</code> method implemented on <code>Human</code> directly.</p>
<p>To call the <code>fly</code> methods from either the <code>Pilot</code> trait or the <code>Wizard</code> trait,
we need to use more explicit syntax to specify which <code>fly</code> method we mean.
Listing 20-19 demonstrates this syntax.</p>
<figure class="listing" id="listing-20-19">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024"><span class="boring">trait Pilot {
</span><span class="boring"> fn fly(&amp;self);
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">trait Wizard {
</span><span class="boring"> fn fly(&amp;self);
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">struct Human;
</span><span class="boring">
</span><span class="boring">impl Pilot for Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("This is your captain speaking.");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Wizard for Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("Up!");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Human {
</span><span class="boring"> fn fly(&amp;self) {
</span><span class="boring"> println!("*waving arms furiously*");
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>fn main() {
let person = Human;
Pilot::fly(&amp;person);
Wizard::fly(&amp;person);
person.fly();
}</code></pre>
<figcaption><a href="#listing-20-19">Listing 20-19</a>: Specifying which traits <code>fly</code> method we want to call</figcaption>
</figure>
<p>Specifying the trait name before the method name clarifies to Rust which
implementation of <code>fly</code> we want to call. We could also write
<code>Human::fly(&amp;person)</code>, which is equivalent to the <code>person.fly()</code> that we used
in Listing 20-19, but this is a bit longer to write if we dont need to
disambiguate.</p>
<p>Running this code prints the following:</p>
<pre><code class="language-console">$ cargo run
Compiling traits-example v0.1.0 (file:///projects/traits-example)
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.46s
Running `target/debug/traits-example`
This is your captain speaking.
Up!
*waving arms furiously*
</code></pre>
<p>Because the <code>fly</code> method takes a <code>self</code> parameter, if we had two <em>types</em> that
both implement one <em>trait</em>, Rust could figure out which implementation of a
trait to use based on the type of <code>self</code>.</p>
<p>However, associated functions that are not methods dont have a <code>self</code>
parameter. When there are multiple types or traits that define non-method
functions with the same function name, Rust doesnt always know which type you
mean unless you use fully qualified syntax. For example, in Listing 20-20, we
create a trait for an animal shelter that wants to name all baby dogs Spot. We
make an <code>Animal</code> trait with an associated non-method function <code>baby_name</code>. The
<code>Animal</code> trait is implemented for the struct <code>Dog</code>, on which we also provide an
associated non-method function <code>baby_name</code> directly.</p>
<figure class="listing" id="listing-20-20">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024">trait Animal {
fn baby_name() -&gt; String;
}
struct Dog;
impl Dog {
fn baby_name() -&gt; String {
String::from("Spot")
}
}
impl Animal for Dog {
fn baby_name() -&gt; String {
String::from("puppy")
}
}
fn main() {
println!("A baby dog is called a {}", Dog::baby_name());
}</code></pre>
<figcaption><a href="#listing-20-20">Listing 20-20</a>: A trait with an associated function and a type with an associated function of the same name that also implements the trait</figcaption>
</figure>
<p>We implement the code for naming all puppies Spot in the <code>baby_name</code> associated
function that is defined on <code>Dog</code>. The <code>Dog</code> type also implements the trait
<code>Animal</code>, which describes characteristics that all animals have. Baby dogs are
called puppies, and that is expressed in the implementation of the <code>Animal</code>
trait on <code>Dog</code> in the <code>baby_name</code> function associated with the <code>Animal</code> trait.</p>
<p>In <code>main</code>, we call the <code>Dog::baby_name</code> function, which calls the associated
function defined on <code>Dog</code> directly. This code prints the following:</p>
<pre><code class="language-console">$ cargo run
Compiling traits-example v0.1.0 (file:///projects/traits-example)
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.54s
Running `target/debug/traits-example`
A baby dog is called a Spot
</code></pre>
<p>This output isnt what we wanted. We want to call the <code>baby_name</code> function that
is part of the <code>Animal</code> trait that we implemented on <code>Dog</code> so that the code
prints <code>A baby dog is called a puppy</code>. The technique of specifying the trait
name that we used in Listing 20-19 doesnt help here; if we change <code>main</code> to
the code in Listing 20-21, well get a compilation error.</p>
<figure class="listing" id="listing-20-21">
<span class="file-name">Filename: src/main.rs</span>
<pre><code class="language-rust ignore does_not_compile"><span class="boring">trait Animal {
</span><span class="boring"> fn baby_name() -&gt; String;
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">struct Dog;
</span><span class="boring">
</span><span class="boring">impl Dog {
</span><span class="boring"> fn baby_name() -&gt; String {
</span><span class="boring"> String::from("Spot")
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Animal for Dog {
</span><span class="boring"> fn baby_name() -&gt; String {
</span><span class="boring"> String::from("puppy")
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>fn main() {
println!("A baby dog is called a {}", Animal::baby_name());
}</code></pre>
<figcaption><a href="#listing-20-21">Listing 20-21</a>: Attempting to call the <code>baby_name</code> function from the <code>Animal</code> trait, but Rust doesnt know which implementation to use</figcaption>
</figure>
<p>Because <code>Animal::baby_name</code> doesnt have a <code>self</code> parameter, and there could be
other types that implement the <code>Animal</code> trait, Rust cant figure out which
implementation of <code>Animal::baby_name</code> we want. Well get this compiler error:</p>
<pre><code class="language-console">$ cargo run
Compiling traits-example v0.1.0 (file:///projects/traits-example)
error[E0790]: cannot call associated function on trait without specifying the corresponding `impl` type
--&gt; src/main.rs:20:43
|
2 | fn baby_name() -&gt; String;
| ------------------------- `Animal::baby_name` defined here
...
20 | println!("A baby dog is called a {}", Animal::baby_name());
| ^^^^^^^^^^^^^^^^^^^ cannot call associated function of trait
|
help: use the fully-qualified path to the only available implementation
|
20 | println!("A baby dog is called a {}", &lt;Dog as Animal&gt;::baby_name());
| +++++++ +
For more information about this error, try `rustc --explain E0790`.
error: could not compile `traits-example` (bin "traits-example") due to 1 previous error
</code></pre>
<p>To disambiguate and tell Rust that we want to use the implementation of
<code>Animal</code> for <code>Dog</code> as opposed to the implementation of <code>Animal</code> for some other
type, we need to use fully qualified syntax. Listing 20-22 demonstrates how to
use fully qualified syntax.</p>
<figure class="listing" id="listing-20-22">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024"><span class="boring">trait Animal {
</span><span class="boring"> fn baby_name() -&gt; String;
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">struct Dog;
</span><span class="boring">
</span><span class="boring">impl Dog {
</span><span class="boring"> fn baby_name() -&gt; String {
</span><span class="boring"> String::from("Spot")
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl Animal for Dog {
</span><span class="boring"> fn baby_name() -&gt; String {
</span><span class="boring"> String::from("puppy")
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>fn main() {
println!("A baby dog is called a {}", &lt;Dog as Animal&gt;::baby_name());
}</code></pre>
<figcaption><a href="#listing-20-22">Listing 20-22</a>: Using fully qualified syntax to specify that we want to call the <code>baby_name</code> function from the <code>Animal</code> trait as implemented on <code>Dog</code></figcaption>
</figure>
<p>Were providing Rust with a type annotation within the angle brackets, which
indicates we want to call the <code>baby_name</code> method from the <code>Animal</code> trait as
implemented on <code>Dog</code> by saying that we want to treat the <code>Dog</code> type as an
<code>Animal</code> for this function call. This code will now print what we want:</p>
<pre><code class="language-console">$ cargo run
Compiling traits-example v0.1.0 (file:///projects/traits-example)
Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.48s
Running `target/debug/traits-example`
A baby dog is called a puppy
</code></pre>
<p>In general, fully qualified syntax is defined as follows:</p>
<pre><code class="language-rust ignore">&lt;Type as Trait&gt;::function(receiver_if_method, next_arg, ...);</code></pre>
<p>For associated functions that arent methods, there would not be a <code>receiver</code>:
There would only be the list of other arguments. You could use fully qualified
syntax everywhere that you call functions or methods. However, youre allowed
to omit any part of this syntax that Rust can figure out from other information
in the program. You only need to use this more verbose syntax in cases where
there are multiple implementations that use the same name and Rust needs help
to identify which implementation you want to call.</p>
<!-- Old headings. Do not remove or links may break. -->
<p><a id="using-supertraits-to-require-one-traits-functionality-within-another-trait"></a></p>
<h3 id="using-supertraits"><a class="header" href="#using-supertraits">Using Supertraits</a></h3>
<p>Sometimes you might write a trait definition that depends on another trait: For
a type to implement the first trait, you want to require that type to also
implement the second trait. You would do this so that your trait definition can
make use of the associated items of the second trait. The trait your trait
definition is relying on is called a <em>supertrait</em> of your trait.</p>
<p>For example, lets say we want to make an <code>OutlinePrint</code> trait with an
<code>outline_print</code> method that will print a given value formatted so that its
framed in asterisks. That is, given a <code>Point</code> struct that implements the
standard library trait <code>Display</code> to result in <code>(x, y)</code>, when we call
<code>outline_print</code> on a <code>Point</code> instance that has <code>1</code> for <code>x</code> and <code>3</code> for <code>y</code>, it
should print the following:</p>
<pre><code class="language-text">**********
* *
* (1, 3) *
* *
**********
</code></pre>
<p>In the implementation of the <code>outline_print</code> method, we want to use the
<code>Display</code> traits functionality. Therefore, we need to specify that the
<code>OutlinePrint</code> trait will work only for types that also implement <code>Display</code> and
provide the functionality that <code>OutlinePrint</code> needs. We can do that in the
trait definition by specifying <code>OutlinePrint: Display</code>. This technique is
similar to adding a trait bound to the trait. Listing 20-23 shows an
implementation of the <code>OutlinePrint</code> trait.</p>
<figure class="listing" id="listing-20-23">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024">use std::fmt;
trait OutlinePrint: fmt::Display {
fn outline_print(&amp;self) {
let output = self.to_string();
let len = output.len();
println!("{}", "*".repeat(len + 4));
println!("*{}*", " ".repeat(len + 2));
println!("* {output} *");
println!("*{}*", " ".repeat(len + 2));
println!("{}", "*".repeat(len + 4));
}
}
<span class="boring">
</span><span class="boring">fn main() {}</span></code></pre>
<figcaption><a href="#listing-20-23">Listing 20-23</a>: Implementing the <code>OutlinePrint</code> trait that requires the functionality from <code>Display</code></figcaption>
</figure>
<p>Because weve specified that <code>OutlinePrint</code> requires the <code>Display</code> trait, we
can use the <code>to_string</code> function that is automatically implemented for any type
that implements <code>Display</code>. If we tried to use <code>to_string</code> without adding a
colon and specifying the <code>Display</code> trait after the trait name, wed get an
error saying that no method named <code>to_string</code> was found for the type <code>&amp;Self</code> in
the current scope.</p>
<p>Lets see what happens when we try to implement <code>OutlinePrint</code> on a type that
doesnt implement <code>Display</code>, such as the <code>Point</code> struct:</p>
<figure class="listing">
<span class="file-name">Filename: src/main.rs</span>
<pre><code class="language-rust ignore does_not_compile"><span class="boring">use std::fmt;
</span><span class="boring">
</span><span class="boring">trait OutlinePrint: fmt::Display {
</span><span class="boring"> fn outline_print(&amp;self) {
</span><span class="boring"> let output = self.to_string();
</span><span class="boring"> let len = output.len();
</span><span class="boring"> println!("{}", "*".repeat(len + 4));
</span><span class="boring"> println!("*{}*", " ".repeat(len + 2));
</span><span class="boring"> println!("* {output} *");
</span><span class="boring"> println!("*{}*", " ".repeat(len + 2));
</span><span class="boring"> println!("{}", "*".repeat(len + 4));
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span>struct Point {
x: i32,
y: i32,
}
impl OutlinePrint for Point {}
<span class="boring">
</span><span class="boring">fn main() {
</span><span class="boring"> let p = Point { x: 1, y: 3 };
</span><span class="boring"> p.outline_print();
</span><span class="boring">}</span></code></pre>
</figure>
<p>We get an error saying that <code>Display</code> is required but not implemented:</p>
<pre><code class="language-console">$ cargo run
Compiling traits-example v0.1.0 (file:///projects/traits-example)
error[E0277]: `Point` doesn't implement `std::fmt::Display`
--&gt; src/main.rs:20:23
|
20 | impl OutlinePrint for Point {}
| ^^^^^ the trait `std::fmt::Display` is not implemented for `Point`
|
note: required by a bound in `OutlinePrint`
--&gt; src/main.rs:3:21
|
3 | trait OutlinePrint: fmt::Display {
| ^^^^^^^^^^^^ required by this bound in `OutlinePrint`
error[E0277]: `Point` doesn't implement `std::fmt::Display`
--&gt; src/main.rs:24:7
|
24 | p.outline_print();
| ^^^^^^^^^^^^^ the trait `std::fmt::Display` is not implemented for `Point`
|
note: required by a bound in `OutlinePrint::outline_print`
--&gt; src/main.rs:3:21
|
3 | trait OutlinePrint: fmt::Display {
| ^^^^^^^^^^^^ required by this bound in `OutlinePrint::outline_print`
4 | fn outline_print(&amp;self) {
| ------------- required by a bound in this associated function
For more information about this error, try `rustc --explain E0277`.
error: could not compile `traits-example` (bin "traits-example") due to 2 previous errors
</code></pre>
<p>To fix this, we implement <code>Display</code> on <code>Point</code> and satisfy the constraint that
<code>OutlinePrint</code> requires, like so:</p>
<figure class="listing">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024"><span class="boring">trait OutlinePrint: fmt::Display {
</span><span class="boring"> fn outline_print(&amp;self) {
</span><span class="boring"> let output = self.to_string();
</span><span class="boring"> let len = output.len();
</span><span class="boring"> println!("{}", "*".repeat(len + 4));
</span><span class="boring"> println!("*{}*", " ".repeat(len + 2));
</span><span class="boring"> println!("* {output} *");
</span><span class="boring"> println!("*{}*", " ".repeat(len + 2));
</span><span class="boring"> println!("{}", "*".repeat(len + 4));
</span><span class="boring"> }
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">struct Point {
</span><span class="boring"> x: i32,
</span><span class="boring"> y: i32,
</span><span class="boring">}
</span><span class="boring">
</span><span class="boring">impl OutlinePrint for Point {}
</span><span class="boring">
</span>use std::fmt;
impl fmt::Display for Point {
fn fmt(&amp;self, f: &amp;mut fmt::Formatter) -&gt; fmt::Result {
write!(f, "({}, {})", self.x, self.y)
}
}
<span class="boring">
</span><span class="boring">fn main() {
</span><span class="boring"> let p = Point { x: 1, y: 3 };
</span><span class="boring"> p.outline_print();
</span><span class="boring">}</span></code></pre>
</figure>
<p>Then, implementing the <code>OutlinePrint</code> trait on <code>Point</code> will compile
successfully, and we can call <code>outline_print</code> on a <code>Point</code> instance to display
it within an outline of asterisks.</p>
<!-- Old headings. Do not remove or links may break. -->
<p><a id="using-the-newtype-pattern-to-implement-external-traits-on-external-types"></a>
<a id="using-the-newtype-pattern-to-implement-external-traits"></a></p>
<h3 id="implementing-external-traits-with-the-newtype-pattern"><a class="header" href="#implementing-external-traits-with-the-newtype-pattern">Implementing External Traits with the Newtype Pattern</a></h3>
<p>In the <a href="../ch10/ch10-02-traits.html#implementing-a-trait-on-a-type">“Implementing a Trait on a Type”</a><!--
ignore --> section in Chapter 10, we mentioned the orphan rule that states
were only allowed to implement a trait on a type if either the trait or the
type, or both, are local to our crate. Its possible to get around this
restriction using the newtype pattern, which involves creating a new type in a
tuple struct. (We covered tuple structs in the <a href="../ch05/ch05-01-defining-structs.html#creating-different-types-with-tuple-structs">“Creating Different Types with
Tuple Structs”</a><!-- ignore --> section in Chapter 5.) The tuple
struct will have one field and be a thin wrapper around the type for which we
want to implement a trait. Then, the wrapper type is local to our crate, and we
can implement the trait on the wrapper. <em>Newtype</em> is a term that originates
from the Haskell programming language. There is no runtime performance penalty
for using this pattern, and the wrapper type is elided at compile time.</p>
<p>As an example, lets say we want to implement <code>Display</code> on <code>Vec&lt;T&gt;</code>, which the
orphan rule prevents us from doing directly because the <code>Display</code> trait and the
<code>Vec&lt;T&gt;</code> type are defined outside our crate. We can make a <code>Wrapper</code> struct
that holds an instance of <code>Vec&lt;T&gt;</code>; then, we can implement <code>Display</code> on
<code>Wrapper</code> and use the <code>Vec&lt;T&gt;</code> value, as shown in Listing 20-24.</p>
<figure class="listing" id="listing-20-24">
<span class="file-name">Filename: src/main.rs</span>
<pre class="playground"><code class="language-rust edition2024">use std::fmt;
struct Wrapper(Vec&lt;String&gt;);
impl fmt::Display for Wrapper {
fn fmt(&amp;self, f: &amp;mut fmt::Formatter) -&gt; fmt::Result {
write!(f, "[{}]", self.0.join(", "))
}
}
fn main() {
let w = Wrapper(vec![String::from("hello"), String::from("world")]);
println!("w = {w}");
}</code></pre>
<figcaption><a href="#listing-20-24">Listing 20-24</a>: Creating a <code>Wrapper</code> type around <code>Vec&lt;String&gt;</code> to implement <code>Display</code></figcaption>
</figure>
<p>The implementation of <code>Display</code> uses <code>self.0</code> to access the inner <code>Vec&lt;T&gt;</code>
because <code>Wrapper</code> is a tuple struct and <code>Vec&lt;T&gt;</code> is the item at index 0 in the
tuple. Then, we can use the functionality of the <code>Display</code> trait on <code>Wrapper</code>.</p>
<p>The downside of using this technique is that <code>Wrapper</code> is a new type, so it
doesnt have the methods of the value its holding. We would have to implement
all the methods of <code>Vec&lt;T&gt;</code> directly on <code>Wrapper</code> such that the methods
delegate to <code>self.0</code>, which would allow us to treat <code>Wrapper</code> exactly like a
<code>Vec&lt;T&gt;</code>. If we wanted the new type to have every method the inner type has,
implementing the <code>Deref</code> trait on the <code>Wrapper</code> to return the inner type would
be a solution (we discussed implementing the <code>Deref</code> trait in the <a href="../ch15/ch15-02-deref.html#treating-smart-pointers-like-regular-references">“Treating
Smart Pointers Like Regular References”</a><!-- ignore -->
section in Chapter 15). If we didnt want the <code>Wrapper</code> type to have all the
methods of the inner type—for example, to restrict the <code>Wrapper</code> types
behavior—we would have to implement just the methods we do want manually.</p>
<p>This newtype pattern is also useful even when traits are not involved. Lets
switch focus and look at some advanced ways to interact with Rusts type system.</p>
</body>
</html>