开发者

Union in c++ are they feasible

开发者 https://www.devze.com 2023-02-16 11:01 出处:网络
C开发者_高级运维an a union in C++ have a member function? How do union with data members and member functions exist if an object is created?

C开发者_高级运维an a union in C++ have a member function? How do union with data members and member functions exist if an object is created?

If I suppose yes, then are they feasible any where. If yes then where?


9.5/1

A union can have member functions (including constructors and destructors), but not virtual (10.3) functions. A union shall not have base classes. A union shall not be used as a base class. An object of a class with a non-trivial constructor (12.1), a non-trivial copy constructor (12.8), a non-trivial destructor (12.4), or a non-trivial copy assignment operator (13.5.3, 12.8) cannot be a member of a union, nor can an array of such objects

What do you mean by How do union with data members and member functions exist if an object is created? Member functions (non-virtual) take no space in an instance of any class/union.


You can also make a template union :

template <typename T>
union Foo {
public:
  Foo() {}
  Foo(const T& value) : _val(value) {}

  const char* data() const {
    return _tab;
  }

  std::size_t size() const {
    return sizeof(T);
  }

  char operator[](unsigned int index) const {
      return _tab[index];
  }

private:
  T _val;
  char _tab[sizeof(T)];
}


The union is a C-structure, and does not work well with C++ types (there are a number of caveats actually). However there already exist a C++ equivalent, which effectively work with all C++ classes and user-defined classes and is even safer than the union!

Behold Boost.Variant!

You can define a boost::variant<std::string, Foo, char> and it'll make sure:

  • that the appropriate constructor/destructor/assignment operator is run, when required
  • that you only access the lastest value that was set

And it even comes with the excellent: boost::static_visitor<Result> which let's you apply a method on the union regardless of its type and provide compile-time checking to warn you whenever you have forgotten one of the possible types!

class MyVisitor: boost::static_visitor<int>
{
public:
  int operator()(std::string const& s) const {
    return boost::lexical_cast<int>(s);
  }

  int operator()(Foo const& f) const { return f.getAsInt(); }

  int operator()(char c) const { return c; }
};

typedef boost::variant<std::string, Foo, char> MyVariant;

int main(int argc, char* argv[]) {
  MyVariant v; // the std::string is constructed

  if (argc % 2) { v = Foo(4); }
  if (argc % 3) { v = argv[1][0]; }
  if (argc % 5) { v = argv[1]; }

  std::cout << boost::apply_visitor(MyVisitor(), v) << '\n';
  return 0;
}

Also... it's as efficient (fast) as a union, and does not involve any dynamic look-up like Boost.Any would.


I don't know if it's valid. Codepad accepts, runs, and gives the expected output from this program

union x {
  int t;
  int k() { return 42;};
};

int main() {
  x y;
  y.t = y.k();
  std::cout << y.t << std::endl;
}


I just added some more things to @maraguida example. I wrote it as a response just to heve more room. It illustrates that not only member functions, but also static member functions and operators can be added.

#include    <iostream>

union x
 {
    int     t;
    float   f;

    int k( )        { return t * 42;};
    static int static_k( )  { return 42;};

    float k_f( )    { return f * 42.0f;};

    unsigned char operator []( unsigned int );
};

unsigned char x::operator []( unsigned int i )
{
    if ( i >= sizeof( x ) )
        return  0;

    return  ( ( unsigned char * )&t )[ i ];
}

int main( )
{
    x   y;
    y.t = x::static_k( );

    std::cout << "y.t\t= " << y.t << std::endl;
    std::cout << "y.f\t= " << y.f << std::endl;
    std::cout << "y.k( )\t= " << y.k( ) << std::endl;
    std::cout << "x::static_k( )\t= " << x::static_k( ) << std::endl;
    std::cout << "y.k_f( )\t= " << y.k_f( ) << std::endl;

    std::cout << "y[ 0 ]\t= " << ( unsigned int )y[ 0 ] << std::endl;
    std::cout << "y[ 1 ]\t= " << ( unsigned int )y[ 1 ] << std::endl;
    std::cout << "y[ 2 ]\t= " << ( unsigned int )y[ 2 ] << std::endl;
    std::cout << "y[ 3 ]\t= " << ( unsigned int )y[ 3 ] << std::endl;
}

It can be compiled with: g++ -Wall union_func.cpp -o union_func

The output is:

$ ./union_func 
y.t     = 42
y.f     = 5.88545e-44
y.k( )  = 1764
x::static_k( )  = 42
y.k_f( )        = 2.47189e-42
y[ 0 ]  = 42
y[ 1 ]  = 0
y[ 2 ]  = 0
y[ 3 ]  = 0

You can, for example, put a conversion operator to another type of your need, if it make sense to your need.

0

精彩评论

暂无评论...
验证码 换一张
取 消