Return-type based dispatch

栏目: IT技术 · 发布时间: 6年前

内容简介:One surprising feature of type inference in languages like Rust is defining functions with generic return types. The idea is that by specifying at some later point in the code which type you want your function to return, the compiler can go back and fill i

One surprising feature of type inference in languages like Rust is defining functions with generic return types. The idea is that by specifying at some later point in the code which type you want your function to return, the compiler can go back and fill in the blanks.

For example, let’s have a look at this function:

fn new<T: Default>() -> T {
  T::default()
}

You pick the output

It has no value parameters, but one type parameter, T . That T is its return type and also used in the function body. You can call it like so:

let x: u32 = new();

Or, being explicit about the type parameter, like this:

let x = new::<i32>();

This is quite neat!

More generic: collect

A promising way to be more generic in Rust is to use more traits! Have a look at how the Iterator::collect method is defined:

fn collect<B: FromIterator<Self::Item>>(self) -> B // ...

You can read this type signature as

Consume self and return something of a type that implements can be made From [an] Iterator for the type of items we are iterating over.

Like above, we call this by specifying what kind of output type want. [Looking][ FromIterator implementors] at some of the types FromIterator is implemented for is pretty revealing of the use cases. You can get:

  • a Vec by collecting any items,
  • a BTreeMap or HashMap by collecting tuples,
  • but also PathBuf by collecting Path s,
  • and String for strings and string slices.

Note: All these types are what you might call “container” types.

One more for the road

More generic? More traits.

There is one more gem hidden in FromIterator :

impl<A, E, V> FromIterator<Result<A, E>> for Result<V, E> where
    V: FromIterator<A>, // ...

This means: You can construct a Result containing any type of container of items A by collecting items that are Result s of type A . (The first Err will make the outer Result be an Err .) Here’s an example, see the docs for another one:

let input: Vec<Result<i32, ()>> = vec![Ok(1), Ok(2)];
let output: Result<Vec<i32>, ()> = input.into_iter().collect();

Note: If you like type theory: What we’re building is a Result<<T<A>, E>> by collecting Result<A, E> s and specifying T .


以上所述就是小编给大家介绍的《Return-type based dispatch》,希望对大家有所帮助,如果大家有任何疑问请给我留言,小编会及时回复大家的。在此也非常感谢大家对 码农网 的支持!

查看所有标签

本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们。

智能时代

智能时代

吴军 / 中信出版集团 / 2016-8 / 68.00

大数据和机器智能的出现,对我们的技术发展、商业和社会都会产生重大的影响。作者吴军在《智能时代:大数据与智能革命重新定义未来》中指出,首先,我们在过去认为非常难以解决的问题,会因为大数据和机器智能的使用而迎刃而解,比如解决癌症个性化治疗的难题。同时,大数据和机器智能还会彻底改变未来的商业模式,很多传统的行业都将采用智能技术实现升级换代,同时改变原有的商业模式。大数据和机器智能对于未来社会的影响是全方......一起来看看 《智能时代》 这本书的介绍吧!

SHA 加密
SHA 加密

SHA 加密工具

XML 在线格式化
XML 在线格式化

在线 XML 格式化压缩工具

正则表达式在线测试
正则表达式在线测试

正则表达式在线测试