std::multimap<A,B>::iterator

[Sang-drax]

OK, I'm tired and it's late, but never mind:

I have:

using namespace std;

multimap<string, Function> Functions;
...
multimap<string, Function>::iterator ifunc;
for (ifunc = Functions.begin(); ifunc < Functions.end(); ++ifunc)
{
...
}

This generates one error of unbelivable length (you know STL errors...)

When I change the < to != it works, but shouldn't < work as well?
Is multimap special?

[OneStiffRod]

What value does Functions.end() return? a number or Bool or other??

[Sang-drax]

it returns multimap<string, Function>::iterator as defined by the standard.

[Sebastiani]

Hmm, perhaps you could overload that one yourself? How would you define .end() -1 though?

[*ClownPimp*]

>When I change the < to != it works, but shouldn't < work as well?

No. "<, >" will only work when the data is stored contiguously in memory (like in an array), but since a map might be implemented with a graph of some sort, each node can be in different locations in memory, not one-after-the-other, hence it is useless to compare the address with the less-than and greater-than operators

[Sang-drax]

OK, that must be the explanation.

But, since ++ can be defined, why not < ?

[elad]

++ and -- operators for the iterator class used with map/multimaps are probably implemented by using an embedded pointer, like with nodes used in a list where, conceivably at least---I've never tried it---you could go like this:

struct node
{
   //data
   node * next;
   void operator++;
};

void node::operator++
{
   this = this->next;
}

I know, the code above is probably full of bugs, but the process should work, once the details are worked out. The added wrinkle here would be that rather than a list, there is probably a tree underlying map/multimap somewhere.

Thus, ++ and -- are only concerned with the next or previous node relative to current node, not some node that may be x number of nodes away from current node, which is what < or > implies.

[Sang-drax]

Of course.

I guess I'm still tired.