How do I initialize a stl vector of objects who th

问题:

suppose I have the following class:

class MyInteger {
private:
  int n_;
public:
  MyInteger(int n) : n_(n) {};
  // MORE STUFF
};

And suppose this class don't have a default trivial constructor MyInteger(). I must always supply an int to initialize it for some reason. And then suppose that somewhere in my code I need a vector<MyInteger>. How do I initialize each MyInteger component in this vector<>?

I have two situations (probably the solution is the same, but I'll state them anyway), a normal variable inside a function:

int main(){
    vector<MyInteger> foo(10);  //how do I initialize each 
                                //MyInteger field of this vector? 
    doStuff(foo);
}

and as data in a class:

class MyFunClass {
private:
   vector<MyInteger> myVector;

public:
   MyFunClass(int size, int myIntegerValue) : myVector(size) {}; 
   // what do I put here if I need the 
   // initialization to call MyInteger(myIntegerValue) for all 
   // components of myVector?
};

Is it possible to do it just in the initialization list or must I write the initialization by hand in the MyFunClass(int, int) constructor?

This seems so very basic, and yet I somehow missed it inmy book and can't find in the web.

回答1:

There are many ways to get there. Here are some of them (in no particular order of presence).

Use vector(size_type n, const T& t) constructor. It initializes vector with n copies of t. For example:

#include <vector>

struct MyInt
{
    int value;
    MyInt (int value) : value (value) {}
};

struct MyStuff
{
    std::vector<MyInt> values;

    MyStuff () : values (10, MyInt (20))
    {
    }
};

Push elements into vector one by one. This might be useful when values should be different. For example:

#include <vector>

struct MyInt
{
    int value;
    MyInt (int value) : value (value) {}
};

struct MyStuff
{
    std::vector<MyInt> values;

    MyStuff () : values ()
    {
        values.reserve (10); // Reserve memory not to allocate it 10 times...
        for (int i = 0; i < 10; ++i)
        {
            values.push_back (MyInt (i));
        }
    }
};

Another option is constructor initialization list, if C++0x is an option:

#include <vector>

struct MyInt
{
    int value;
    MyInt (int value) : value (value) {}
};

struct MyStuff
{
    std::vector<MyInt> values;

    MyStuff () : values ({ MyInt (1), MyInt (2), MyInt (3) /* ... */})
    {
    }
};

Of course, there is an option to provide default constructor and/or use something other than std::vector.

Hope it helps.

回答2:

If the elements of the vector are not default-constructible, then there are certain things you cannot do with the vector. You cannot write this (example 1):

vector<MyInteger> foo(10);

You can, however, write this (example 2):

vector<MyInteger> foo(10, MyInteger(37));

(This only requires a copy constructor.) The second argument is an initializer for the elements of the vector.

In your case, you could also write:

vector<MyInteger> foo(10, 37);

...since MyInteger has a non-explicit constructor taking "int" as argument. So the compiler will cast 37 to MyInteger(37) and give the same result as example 2.

You might want to study the documentation on std::vector.

回答3:

vector<MyInteger> foo(10, MyInteger(MY_INT_VALUE));

MyFunClass(int size, int myIntegerValue) : myVector(size, MyInteger(myIntegerValue)) {}; 

回答4:

Besides all answers which answered the question very well, in a case that your class MyInteger is not copy-constructible, you could use this trick : instead of creating vector< MyInteger>, you could create vector< shared_ptr< MyInteger > >

回答5:

Initialization lists can be used without reference to the underlying objects.

#include <string>
#include <vector>
using namespace std;


class Test
{
   public:
   struct NumStr
   {
      int num;
      string str;
   };

   Test(vector<int> v1,vector<NumStr> v2) : _v1(v1),_v2(v2) {}
   vector<int> _v1;
   vector<NumStr> _v2;
};

int main()
{
   Test t={ {1,2,3}, {{1,"one"}, {2,"two"}, {3,"three"}} };
   cout << t._v1[1] << " " << t._v2[1].num << " " << t._v2[1].str << endl;
   return 0;
}

output: 2 2 two