De-referencing the last iterator position

[dwsmith]

Suppose I have a vector<int> v1 = {some set of integers} and I want to add the 0 and the last index, 1 and the last - 1 index, and so on. How can de-reference the last index?

for (auto it = v1.begin(); it != v1.begin() + (v1.end() - v1.begin())/2;      
       ++it) {                                                                  
    cout << "The sum of indices " << it - v1.begin() << " and "                 
         << v1.end() - it << " is ";         
    // << *it + *(v1.end() - it) << endl;
}

The following code

v1.end() - it

prints out the last index, but I can't figure out how to de-reference it; therefore, I am unable to add the values for the specified indices.

[JLBorges]

#include <iostream>
#include <vector>

int main()
{
    std::vector<int> seq { 12, 34, 67, 23, 82, 38, 77, 52, 26 } ;

    for( int v : seq ) std::cout << v << ' ' ;
    std::cout << '\n' ;

    // option one: iterate in reverse
    int delta = -1 ; // ++delta == 0
    for( auto iter = seq.rbegin() ; iter != seq.rend() ; ++iter )
        *iter += ++delta ;
    for( int v : seq ) std::cout << v << ' ' ;
    std::cout << '\n' ;

    // option two: iterate forward
    delta = seq.size() ; // --delta == seq.size() - 1
    for( int& v : seq ) v += --delta ;
    for( int v : seq ) std::cout << v << ' ' ;
    std::cout << '\n' ;
}

http://coliru.stacked-crooked.com/a/791f9fd8473aa864

[dwsmith]

I have made progress but haven't yet achieved the desired results.

  auto j = v1.end() - v1.begin() - 1;                                           
  for (auto it = v1.begin(); it != v1.begin() + (v1.end() - v1.begin())/2;      
       ++it) {                                                                  
    cout << "The sum of indices " << it - v1.begin() << " and "                 
         << v1.end() - it << " is " << *it + *(j - it) << endl;                 
  }   

I need to subtract j from it, but if I do, I am met with:

no match for ‘operator-’ (operand types are
‘long int’ and ‘__gnu_cxx::__normal_iterator<int*, std::vector<int> >’)
          << v1.end() - it << " is " << *it + *(j - it) << endl;

If I subtract it from j, there are no errors but this isn't correct. How can the operators not match when it is j - it but be okay when it is it - j?

[JLBorges]

> How can the operators not match when it is j - it but be okay when it is it - j?

Iterators have pointer like semantics.

int main()
{
    int a[20]{} ;

    int* it = a + 10 ;

    *(it-5) == 8 ; // fine: a[10-5] == 8 ;
    // for pointer addition, any one of the two operands is a pointer type
    // and the other operand is a numeric type

    *(15-it) == 0 ; // *** error: invalid operands to binary -
    // for pointer subtraction,
    // the  left operand must be a pointer type and the right operand a numeric type
    // (or both operands must be pointers)
}

[dwsmith]

@JlBorges, this is the output from the GCC compiler when I compile the code:

error: no match for ‘operator-’ (operand types are
‘long int’ and ‘__gnu_cxx::__normal_iterator<int*, std::vector<int> >’)
          << v1.end() - it << " is " << *it + *(j - it) << endl;

but since I defined j as

auto j = v1.end() - v1.begin - 1;,

j will be larger than it. I want to add the last item and the first, second to last and the second, and so on. With that being the case, j - it > 0 and it - j < 0.

[JLBorges]

> j will be larger than it.

j is not an iterator; it is a number of an integral type
(typically std::ptrdiff_t; in the GCC implementation above, this is an alias for long int)

auto j = v1.end() - v1.begin - 1; is equivalent to auto j = v1.size() - 1 ;

For pointer subtraction, the left operand must be of a pointer type.
For iterator subtraction, the left operand must be of an iterator type.