Thread: Whoa, are reverse iterators slower and should I really care?

What was it, I'm using these reverse iterators to add nodes from a graph to a stack for iterative traversal. My real code looks like this :

Code:

void search_tree(tree *root, particle *p, vector<tree*> & leaves) {


   set<tree*> visited;


   stack<tree*> traversal;
   traversal.push(root);


   while (!traversal.empty()) {


      tree *curr = traversal.top();


      if (!visited.insert(curr).second) {


         traversal.pop();
         continue;
      }


      if (curr->del == 0) {


         if (inside_tetra(curr->t, p) == -1) {


            traversal.pop();
            continue;
         }


         if (curr->children.empty()) {


            leaves.push_back(curr);
            traversal.pop();
            continue;
         } else {


            //for (auto it = curr->children.end(); it < curr->children.end(); ++it)
               //traversal.push(*it);


            for (auto it = curr->children.crbegin(); it != curr->children.crend(); ++it)
               traversal.push(*it);


            continue;
         }
      } else if (curr->del == 1) {


         //for (auto it = curr->children.begin(); it < curr->children.end(); ++it)
            //traversal.push(*it);


         for (auto it = curr->children.crbegin(); it != curr->children.crend(); ++it)
            traversal.push(*it);


         continue;
      }
   }


   return;
}

I only decided to add the children in reverse order because the way my code works, the way in which children are processed matters for the output and I was trying to make sure my iterative solution matches exactly to my recursive implementation which I've been using as a reference. The outputs match though. I was just trying to optimize the code.