fix iterator bugs in path optimizers

[Kazfyx]

In the simple optimizer and advanced optimizer, some iterators are increased in the loop body. But at the end of each loop, these iterators will be increased again by the for-loop structure. When the program is running in debug mode, an assertion often occurs that an iterator is behind the end iterator of a VectorList. I think this is not the real intention of the developers, and it would not get a correct optimization result.

[rickertm]

Thank you for reporting this issue and the suggested fixes. This was introduced during the refactoring in 5064ded and is indeed not the intended behavior.

The refactoring to for loops and the updated init statements do not have to be reverted, just the iteration expression, otherwise this fix looks good to me. Would you mind updating this accordingly?

[Kazfyx]

Thank you for reporting this issue and the suggested fixes. This was introduced during the refactoring in 5064ded and is indeed not the intended behavior.

The refactoring to for loops and the updated init statements do not have to be reverted, just the iteration expression, otherwise this fix looks good to me. Would you mind updating this accordingly?

As iterator increasing is fully controlled in the loop body, the iteration expression must be empty. I may write something like this:

for (VectorList::iterator i = path.begin(), j = ::std::next(i), k = ::std::next(j); i != path.end() && j != path.end() && k != path.end(); )
{
    ...
}

This is not a typical usage of for-loop, and I think it has no advantage compared whith a while-loop. Of course, if you insist on for-loop, I can update it.

[rickertm]

The iteration expression should be empty in this case, as in your example. The behavior of the two variants is identical, but the scope of the loop variables is limited in the case of the for loop.