C++ Primer Plus

The vector class has more capabilities than the built-in array type, but this comes at a cost of slightly less efficiency. If all you need is a fixed-size array, it could be advantageous to use the built-in type. However, that has its own costs of lessened convenience and safety. C++11 responded to this situation by adding the array template class, which is part of the std namespace. Like the built-in type, an array object has a fixed size and uses the stack (or else static memory allocation) instead of the free store, so it shares the efficiency of built-in arrays. To this it adds convenience and additional safety. To create an array object, you need to include the array header file. The syntax is a bit different from that for a vector:

#include

...

using namespace std;

array ai;   // create array object of 5 ints

array ad = {1.2, 2.1, 3.43. 4.3};

More general, the following declaration creates an array object arr with n_elem elements of typeName:

array<typeName, n_elem> arr;

Unlike the case for vector, n_elem can’t be a variable.

With C++11, you can use list-initialization with vector and array objects. However, that was not an option with C++98 vector objects.

Comparing Arrays, Vector Objects, and Array Objects

Perhaps the simplest way to understand the similarities and differences between arrays, vector objects, and array objects is to look at a brief example (Listing 4.24) that uses all three approaches.

Listing 4.24. choices.cpp

// choices.cpp -- array variations

#include

#include    // STL C++98

#include     // C++11

int main()

{

    using namespace std;

// C, original C++

    double a1[4] = {1.2, 2.4, 3.6, 4.8};

// C++98 STL

    vector a2(4);   // create vector with 4 elements

// no simple way to initialize in C98

    a2[0] = 1.0/3.0;

    a2[1] = 1.0/5.0;

    a2[2] = 1.0/7.0;

    a2[3] = 1.0/9.0;

// C++11 -- create and initialize array object

    array a3 = {3.14, 2.72, 1.62, 1.41};

    array a4;

    a4 = a3;     // valid for array objects of same size

// use array notation

    cout << "a1[2]: " << a1[2] << " at " << &a1[2] << endl;

    cout << "a2[2]: " << a2[2] << " at " << &a2[2] << endl;

    cout << "a3[2]: " << a3[2] << " at " << &a3[2] << endl;

    cout << "a4[2]: " << a4[2] << " at " << &a4[2] << endl;

// misdeed

    a1[-2] = 20.2;

    cout << "a1[-2]: " << a1[-2] <<" at " << &a1[-2] << endl;

    cout << "a3[2]: " << a3[2] << " at " << &a3[2] << endl;

    cout << "a4[2]: " << a4[2] << " at " << &a4[2] << endl;

    return 0;

}

Here’s some sample output:

a1[2]: 3.6 at 0x28cce8

a2[2]: 0.142857 at 0xca0328

a3[2]: 1.62 at 0x28ccc8

a4[2]: 1.62 at 0x28cca8

a1[-2]: 20.2 at 0x28ccc8

a3[2]: 20.2 at 0x28ccc8

a4[2]: 1.62 at 0x28cca8

Program Notes

First, notice that whether we use a built-in array, a vector object, or an array object, we can use the standard array notation to access individual members. Second, you can see from the addresses that array objects use the same region of memory (the stack, in this case) as the built-in array, whereas the vector object is stored in a different region (the free store, or heap). Third, note that you can assign an array object to another array object. For built-in arrays, you have to copy the data element-by-element.

Next, and this deserves special attention, note this line:

a1[-2] = 20.2;

What does an index of -2 mean? Recall that this translates to the following:

*(a1-2) = 20.2;

Expressing this in words, see where a1 points, move backward two double elements, and put 20.2 there. That is, store the information at a location outside of the array. C++, like C, does not check for such out-of-range errors. In this particular case, that location turned out to be in the array object a3. Another compiler placed the wayward 20.2 in a4, and other compilers might make yet other bad choices. This is an example of the unsafe behavior of built-in arrays.

Do the vector and array objects protect against this behavior? They can if you let them. That is, you still can write unsafe code, such as the following:

a2[-2] = .5;    // still allowed

a3[200] = 1.4;

However, you have alternatives. One is using the at() member function. Just as you can use the getline() member function with the cin object, you can use the at() member function with objects of the vector or array type:

a2.at(1) = 2.3;  // assign 2.3 to a2[1]