Question about deleting pointers

[Wynd]

Say you have this code:

float *Pi = new float;
*Pi = 3.141592;
delete Pi;

Why wouldn't you just use this:

float Pi = 3.141592;

[Zaei]

You could. Using a pointer in that situation is just dumb. A much better example would be:

float* floats = new float[1000];

Now you have, basically, an array of floats. Then you can delete it as above. Another example is:

myStruct* strct = new myStruct;
SomeFunc(strct);

...
where sizeof(myStruct) > 4

in this case, you are passing a pointer(DWORD), instead of the entire struct, which increases speed. You can also create pointers to functions, which help a lot when doing some things (callback functions). Its all in knowing when to use a pointer. In your example, there is just no reason. But other times, is speeds execution time up, by not haveing to copy data from location to location.

Z.

[parksie]

One thing you missed:

float *i = new float;
delete i;

float *j = new float[100];
delete[] j;

[CornedBee]

more reasons: the C++ STL library

#include <vektor>
using namespace std;

class CParentClass
{ // members
}

class CFirstChildClass : public CParentClass
{  // Blabla  
}

class CSecondChildClass : public CParentClass
{  // Whatever
}

void SomeFunc()
{
     CParentClass *pFC = new CFirstChildClass;
     // ask for user input and fill class members
     // insert the pointer into a vektor <CParentClass*> object that is always available
}

void SomeOtherFuncThatIsCalledAtAnEntirelyOtherTime()
{
      // you can now access the object
}

Advantages:
1) Polymorphic access: It doesn't matter which child class the contents of the vektor are.
2) You still have exactly the object you created, not a copy as if you were using vektor<CParentClass>
3) If you used CFirstChildClass obj; and inserted &obj into the vektor, obj would be out of scope in SomeOtherFunc... and most probably deleted.

[parksie]

vector not vektor

You don't need the STL to use virtual inheritance, which is what goes on here:

class Parent {
public:
    virtual void Func() = 0;
};

class ChildA : public Parent {
public:
    void Func() { cout << "Die!" << endl; }
};

class ChildB : public Parent {
public:
    void Func() { cout << "Live!" << endl; }
};

void somecode() {
    Parent *ptr;
    ChildA *thing = new ChildA;
    ChildB *other = new ChildB;

    ptr = thing;
    ptr->Func();

    ptr = other;
    ptr->Func();

    delete thing;
    delete other;
}

[CornedBee]

true, but this way I can at the same time show the "no copy" and the polymorphic advantages.

[parksie]

Okay then I just wouldn't mix templates in if they're still on the "*** do pointers do" stage

Not quibbling, though

[CornedBee]

Yeah, maybe write an essay about the difference between heap and stack?

[parksie]

Aaaargghhh yeah!

Damn that really messed me up when I was learning. Hmm...perhaps another FAQ entry?

[CornedBee]

It's just that I don't have time for it at the moment.

[Wynd]

Originally posted by parksie
Okay then I just wouldn't mix templates in if they're still on the "*** do pointers do" stage

Heh, that's pretty much me. I know the basics of what they do, but I'm still a little fuzzy on it.