STL vector core dump

Hey everybody. I have an issue where pushing a new item onto an STL vector, I get a core dump. Now, the capacity of the vector is not exceeded (its only 3). The vector contains integers. Now, in my application, I push stuff to the vector and works fine until that one point. Basically, I do this:

Code:

---

vector<int> vec;
vec.push_back(newInt);

---

and get a core dump. Now, the function which involves this, is part of the class reached from the outside by a few iterator references (in one instance one of those iterators is a const_iterator, and it is converted into a regular iterator). All the references are valid, and I know that because I put couts before and after the push_back. I see the first cout, but not the second...Ay ideas..help is cool.

Quote:

---

All the references are valid, and I know that because I put couts before and after the push_back.

---

That is no guarantee that the iterators are valid. I suggest that you post the smallest and simplest compilable program that demonstrates the problem.

I realized something else too: if I cout the newInt, it is correct. However, if I try to push it, a core dump happens. Now, here is the kicker: if I replace newInt with some other integer (local variable or simply a constant, the core dump dissapears). Very strage.

Code:

---

void somClass::addNewInt(const int& newInt)
{
    vec.push_back(newInt);
}

---

Yet even stranger:

If I add the following code, the core dump goes away:

Code:

---

void somClass::addNewInt(const int& newInt)
{
    vector<int> vec2;
    vec2.push_back(newInt);

    vec.push_back(newInt);
}

---

Just out of curiosity, are there any calls to erase on this vector at some point?

Quote:

---

Originally Posted by swoopy

Just out of curiosity, are there any calls to erase on this vector at some point?

---

Nope.

Regardless of the existence of iterators, valid or not, a call to push_back() on a vector should never fail. I would suspect you've caused memory corruption somewhere.

Quote:

---

Originally Posted by creativeinspira

I realized something else too: if I cout the newInt, it is correct. However, if I try to push it, a core dump happens. Now, here is the kicker: if I replace newInt with some other integer (local variable or simply a constant, the core dump dissapears). Very strage.

Code:

---

void somClass::addNewInt(const int& newInt)
{
    vec.push_back(newInt);
}

---

---

since newInt is a reference, could you show us where you've called your addNewInt() function from, and the source of newInt - perhaps include the block of code where you've declared 'vec' aswell (I suspect there may be a dangling reference somewhere)

Thanks to you for all who tried to help. I can't post detailed code. I will just figure this out myself.

One difference between one of the working code snippets and the not working code snippet is the declaration of the vector. When it works it is declared locally, but when it fails it is apparaently declared as a member of a class.

One place to look is that class object itself. Oftentimes, you will get this kind of error when the object that the function is called on (in this case a somClass object) is bad. So I would go down the call stack and see if a pointer to somClass is used and make sure it is valid.