std::vector vs. C-style arrays

[Axter]

Originally posted by etaoin
Andreas, thanks for your reply.

Even when we reduce the problem to a vector of a pre-determined size - my question is, how can I initialize the elements without invoking the copy constructor and creating a temporary element? With a C array, we can do the following (assuming, for simplicity, that the members of Element were public):

Element* pElements = new Element[NUM_ELEMENTS];

for(int i = 0; i < NUM_ELEMENTS; i++)
{
  pElements[i].m_time = someTime;
  pElements[i].m_name = someName;
  pElements[i].m_value = someValue;
}

This will simply assign the values once to each element in the array, leading to no overhead as compared to the original code. But how can I do the same thing properly with a std::vector?

You can do this with a vector object just as easy.
Example:

std::vector<Element> pElements(NUM_ELEMENTS);

for(int i = 0; i < NUM_ELEMENTS; ++i)
{
  pElements[i].m_time = someTime;
  pElements[i].m_name = someName;
  pElements[i].m_value = someValue;
} 

Using the same method, you can access the same members and assign the values just as you would with the C-Style array.

Moreover, if you use iterators, you can make this same code faster using the vector method.

std::vector<Element> pElements(NUM_ELEMENTS);

for(std::vector<Element>::iterator i = pElements.begin(), e = pElements.end();
        i!=e;++i)
{
  i->m_time = someTime;
  i->m_name = someName;
  i->m_value = someValue;
} 

The above version using iterators will have a much better performance then you're original code using a C-Style array.

Vector iterators have a faster access time then does index access to a C-Style array.

Almost anything you can do with a C-Style array, you can do with a vector.