Re: list iterators modified only with reverse_iterator

Am 04.03.2011 20:20, schrieb backminator:


I'm not convinced that your example demonstrates that the iterator
values has been "invalidated". To proof that we must test our
expectations with experiments that make these expectations observable.

First we need to start with the definition of a valid iterator. We check
the current standard and find - nothing. At least no clear answer to
that question. There is an issue that is concerned about this lack of
definition, see:

http://www.open-std.org/jtc1/sc22/wg21/docs/lwg-active.html#1213

In absence of such a definition it might help to compare with existing
definitions, e.g. that of the pre-standard library and we can look at
the SGI definition shown in

http://www.sgi.com/tech/stl/InputIterator.html

which says:

"An iterator is valid if it is dereferenceable or past-the-end."

I see nothing within this definition that guarantees that repeated calls
of the expression *it will return the same value when during that time
the underlying container is modified. Thuis I don't think that your
observation that the result of this expression is different with an
intervening mutating container proofs that the iterator has been
invalidated. Looking at the definition of reverse_iterator's operator*
we find that the return value is that of a decrementing a copy of the
current iterator and returning the value of that copy. Since a new value
has been inserted to the point where we are decrementing the copy, we
obtain the expected value after the modification of the container.