非默认构造函数的C ++继承 [英] C++ inheritance with non-default constructors

问题描述

对于一个项目，我正在研究存在以下情况的情况.有一个foo类，在构造时需要一个标识号x.因此，其构造函数参数类似于foo(x).

For a project I am looking at a case where the following situation exists. There is a class foo which requires an identifying number x when constructed. So its constructor argument looks something like foo(x).

现在，我想派生一个基于foo的类，称为bar，该类基本上将在内部容纳两个foo，并且在外部世界中将充当单个foo.我要执行此操作的方法是使用bar的两个构造函数参数y和z，然后将在构造函数中生成foo(y)和foo(z).我希望一切都清楚.

Now I want to derive a class based on foo, called bar, which will basically hold two foo's internally and to the outside world acts as a single foo. The way I want to do this is by use of two constructor arguments y and z for bar, which will then, in the constructor, generate foo(y) and foo(z). I hope it's clear.

我当前的实现类似于:

class bar : public foo {
public:
    bar(int y, int z)
    {
        _foo_ptr_1 = foo::foo(y);
        _foo_ptr_2 = foo::foo(z);
    };

private:
    foo *_foo_ptr_1;
    foo *_foo_ptr_2;
};

其中foo是一个类，它是一个已经存在的类，类似于:

Where foo is a class that is an already existing class looking something like:

class foo {
public:
    foo(int x) :
    _private_var(x) {
    };
};

现在，我知道由于foo没有带有零参数的默认构造函数，因此它将不起作用.由于某些原因，我宁愿不向foo添加零参数的无意义构造函数.有没有一种很好的方法来使此工作?现在，我的编译器(在Ubuntu下为gcc)给我以下错误:

Now, I know that since foo has no default constructor with zero arguments this will not work. I would prefer not to add a meaningless constructor with zero arguments to foo for several reasons. Is there a nice way to get this working? Right now my compiler (gcc under Ubuntu) gives me the following error:

error ::没有匹配函数可调用foo :: foo()

error::no matching function for call to foo::foo()

尝试解决此问题的快速尝试是替换bar的构造函数:

A quick attempt to try and work around this is by replacing the constructor of bar:

bar(int y, int z) : foo(y)

但这也不起作用.

我还无法在线找到一个可行的解决方案，因此非常感谢您的帮助.

I have not been able to find a working solution for this yet online, and would be gratefull for any help.

推荐答案

对我来说这很好:

bar(int y, int z) : foo(y)
{
  _foo_ptr_1 = new foo(y);
  _foo_ptr_2 = new foo(z);
}

请不要试图像静态方法那样直接调用类构造函数并不是在C ++中构造类的方式.至少，您如上所述使用 new classname(arguments).

Please not that trying to invoke a class constructor directly like it was a static method is not how classes are constructed in C++. At the very least, you use new classname(arguments), as above.

这不是真正的建议，但是...使用 new 和 delete 进行显式的内存管理很容易出错.

This is not really recommended however... explicit memory management using new and delete is easy to get wrong.

首先，考虑到您可能根本不必分配任何 foo 实例..您可以只具有简单的成员变量:

First, consider that you may not have to allocate any foo instances at all.. you can just have simple member variables:

class bar : public foo {
public:
  bar(int y, int z) : foo(y), _foo_1(y), _foo_2(z)
  {
  }

private:
  foo _foo_1;
  foo _foo_2;
};

如果您真的必须以困难的方式构造新实例，则应该考虑使用智能指针，例如 unique_ptr .

If you really must construct new instances the hard way, you should consider using smart pointers such as unique_ptr.

class bar : public foo {
public:
  bar(int y, int z) : foo(y), _foo_ptr_1(new foo(y)), _foo_ptr_2(new foo(z))
  {
  }

private:
  std::unique_ptr<foo> _foo_ptr_1;
  std::unique_ptr<foo> _foo_ptr_2;
};

unique_ptr 可确保您的资源在其所有者被销毁时被正确地重新分配.如果您的 _foo 成员变量不完全属于其父 bar 实例，则可以改用 shared_ptr .值得一读这些实用程序类的使用.

unique_ptr ensures that your resources are correctly deallocated when their owner is destroyed. If your _foo member variables don't belong exclusively to their parent bar instance, then you can use shared_ptr instead. It is well worth reading up on the use of these utility classes.

这篇关于非默认构造函数的C ++继承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！