stl std::iterator

[c94wjpn]

If you've got an iterator iter, how do you get an iterator pointing to the previous thing EASILY?

Like eg

while (iter!= (iter0 - 1))

No answers of the form

That's easy, you just declare another iterator:

Iterator iter1;
iter1 = iter;
--iter1;

0 points.

[kuphryn]

What is the significance of using a temporary variable? It will not affect the over performance.

Kuphryn

[Yves M]

What iterator category are you talking about ? Random access iterators (e.g. for std::vector or std::string), bidirectional iterators (e.g. std::list, std::map, std::set) or forward iterators (e.g. istream or ostream) ?

[Graham]

Naughty, Yves - that's a trick question, dropping forward iterators into it, when he's asked for the previous.

[c94wjpn]

Originally posted by Yves M
What iterator category are you talking about ? Random access iterators (e.g. for std::vector or std::string), bidirectional iterators (e.g. std::list, std::map, std::set) or forward iterators (e.g. istream or ostream) ?

Aha I see that random access iterators have that thing (iter-1) which is what I thought you could do. However since I'm using a map, it's not possible. Oh dear. Why isn't (iter-1) defined when you've got an iterator over a map?

[Graham]

It is (sort of) - you just express it as "--iter". std::map::iterator is a bidirectional iterator, so has operator++ and operator-- (both forms of each) defined. Only random access iterators have operator+ and operator- defined.

[c94wjpn]

Originally posted by Graham
It is (sort of) - you just express it as "--iter". std::map::iterator is a bidirectional iterator, so has operator++ and operator-- (both forms of each) defined. Only random access iterators have operator+ and operator- defined.

No, "--iter" and "iter-1" are significantly different because the former alters iter, but iter-1 does not.

--iter is, like iter = iter-1;

[Yves M]

Originally posted by c94wjpn
Aha I see that random access iterators have that thing (iter-1) which is what I thought you could do. However since I'm using a map, it's not possible. Oh dear. Why isn't (iter-1) defined when you've got an iterator over a map?

Well, because making a map's iterator random access would make you incurr big performance penalties. The ++ and -- operators for map's iterators are already amortized constant time, so it's a little slower than constant time. Getting a random access iterator over a map would be amortized linear time (so something like O(N log(log(N))) ).

It just doesn't make sense. Anyways, I don't see the problem in having a temporary iterator which points at the previous element.

[Graham]

Originally posted by c94wjpn
No, "--iter" and "iter-1" are significantly different because the former alters iter, but iter-1 does not.

--iter is, like iter = iter-1;

Which is why I said "sort of": you can get the previous iterator (assuming that iter != cont.begin() ), but that's as far as it goes.

[c94wjpn]

Originally posted by Graham
Which is why I said "sort of": you can get the previous iterator (assuming that iter != cont.begin() ), but that's as far as it goes.

ok thanks.