[STL Unique] what are the duplicated elements?

[rssmps]

Question on what happens to the extra elements that are found to be duplicates. Specifically, the last sentence in the following paragraph.

http://www.sgi.com/tech/stl/unique.html
Every time a consecutive group of duplicate elements appears in the range [first, last), the algorithm unique removes all but the first element. That is, unique returns an iterator new_last such that the range [first, new_last) contains no two consecutive elements that are duplicates. [1] The iterators in the range [new_last, last) are all still dereferenceable, but the elements that they point to are unspecified.

So it's dereferenceable but you get garbage for those elements?

in SGI's example, they created new iterator.
vector<int>::iterator new_end = unique(V.begin(), V.end());

what if I reused the end() iterator?
V.end() = unique(V.begin(), V.end());

is this causing memory leak from the new end() to the old end()?

Let's say there are 10 total elements and 5 are duplicates. So according to the SGI description, the size() will still be 10 after unique. What's the preferred way to get the right size? do you always have to involve an extra erase command after unique?

[Danny]

unique rearranges the container, pushing duplicates past the new logical end, but not erasing them. it returns an iterator that marks the new logical end of the container, i.e., the first non-unique element (which is one position past the last unique element). so if you have the following sequence:
1,2,3,2,5
the rerranged container after calling unique is:
1,2,3,5,[2]
the new_end points to the fifth element (instead of the sixth, as in the origianl container. remmeber that end() always returns one past the last valid element)

The iterators aren't strictly invalidated because the container doesn't reallocate or shrink. However, the values to which the iterators point might change, so if for instance *it was 5, it could now be 2. The values of the dereferenced iterators are unspecified because you can't tell how the duplicates (i.e., the elements at new_end and higher) are ordered. So if you access such an iterator, your code will not crash but the value is unknown, or at least not portably known.
This implies that in order to remove the duplciates you have to call erase (new_end, end) and if you want to release the memory occupied by these elements you have to use the self-swapping idiom described here:
http://www.devx.com/cplus/10MinuteSolution/29484

[rssmps]

OK, I get it.
The swap is a neat trick!

Thanks!