“使用"类作用域的类型别名:[何时]方法中的用法能否优先于类型别名? [英] "using" type alias of class scope: [when] can usages in methods PRECEDE the type alias?

问题描述

昨天，当我能够编译具有使用

Yesterday, I was (pleasantly) surprised when I was able to compile code that had a method which used a using type alias even though the declaration of the alias was not until later in the class definition.

• 这种类型别名的转发"用法是否有效?(我想是这样，因为Clang 5和GCC 4.9都以这种方式工作.)
• 哪些规则涵盖了此行为以及方法主体和方法签名用法之间的区别?

案例1-使用在方法之后声明的方法，在方法体内有效(仅?)

#include <string>
#include <iostream>

struct X {
  std::string create() {     // fails to compile if Y used in signature
      return Y{"hello!"};    // compiles when Y here
  }

  using Y = std::string;     // declared at bottom
};

int main() 
{
    std::cout << X().create() << std::endl;
}

案例2-使用上面声明的签名也[有效]

#include <string>
#include <iostream>

struct X {
  using Y = std::string;     // declared at top

  Y create() {               // can use Y here as well
      return Y{"hello!"};
  }
};

int main() 
{
    std::cout << X().create() << std::endl;
}

推荐答案

这与 完整类上下文 .当您处于成员函数的主体中时，该类将被视为完整的，并且可以使用该类中定义的任何内容，无论在类中的何处声明它.

This has to do with the complete-class context. When you are in the body of a member function, the class is considered complete and can use anything defined in the class, no matter where in the class it is declared.

成员函数的返回类型不属于完整类上下文的一部分，因此您只能使用代码中当时已知的名称.

The return type of a member function is not part of the complete-class context so you can only use names that are known about at that point in the code.

这篇关于“使用"类作用域的类型别名:[何时]方法中的用法能否优先于类型别名?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！