How do I check for a valid node pointer?

[agamemnus]

Hi,

Is there any function that I can use to check whether a node pointer is valid? (ie: points to a complete node)

[ent1ty]

Code: Select all

if(node)
{
    //do your funky stuff
}

off course, when you need a node pointer no more, you have to set it to 0

[CuteAlien]

Code: Select all

if(node)
{
    //do your funky stuff
}

off course, when you need a node pointer no more, you have to set it to 0

And at least as important - you have to initialize all your pointers with 0 yourself. In c++ you usually do that in the constructor - you initialize all pointers of a class (and usually also all other variables) to 0. This is so important because otherwise your pointers will not be initialized at all and have a random value.

[Virion]

in irrlicht, the engine already initialized the node as 0 by default. so if its an invalid node you will get a zero. so you can do this:

Code: Select all

if (node != 0)
{
  //do your stuff
}

[CuteAlien]

in irrlicht, the engine already initialized the node as 0 by default

That's wrong. Actually a software library can't(!) initialize your pointers. How should it do that? Those pointers are in your own code.
Initialize your variables in c++ or they will have random values - it does not matter of which type they are.

edit: What Irrlicht often does is to return 0 as value when some function fails. For example if you call addSceneNode() then it will return 0 on failure and if you assign that to a variable it will have a value afterwards. But do not let that stop you from initializing your pointers - it is basically always a very bad idea not doing that. Not doing so will lead to the strangest effects (like code only running in debug...hehe - I just see your other thread)

[agamemnus]

I can't set it to 0 when it becomes invalid in some cases. That's why I need to be able to test if a pointer is valid, not if it's 0. (test without the program crashing!)

[Bate]

As far as I know there is no other way to check for a valid pointer except for checking the address.
However, since you can't fully control your pointers, it's maybe more about general design?

[Virion]

it is basically always a very bad idea not doing that. Not doing so will lead to the strangest effects (like code only running in debug...hehe - I just see your other thread)

damn. lol

[agamemnus]

As far as I know there is no other way to check for a valid pointer except for checking the address.
However, since you can't fully control your pointers, it's maybe more about general design?

Obviously, yeah... :\

Here's how I end my program (in Freebasic):

irrStop (calls various commands in Irrlicht to stop Irrlicht)
system

The system calls all my destructors... since Irrlicht already freed my data (but didn't clear the pointers), the destructors that destroy Irrlicht data would create a segmentation error because they'll be trying to free data that is already freed... only way to fix this atm is to simply remove the Irrlicht-specific destructor. That just gets a bit confusing, though.. what if I want to destroy an object in the middle of my pogram? I need that destructor to stay there!

If the destructor were to stay, I'd have to manually call all the Irrlicht-specific destructors before calling irrStop. Annoying.

[Acki]

The system calls all my destructors

ehm, you must not call a destructor on your own !!!
they'll be executed automatically when the instance gets deleted !!!
if you want to clear something manually then best is to create a "cleanup" function and call this, and there delete all the pointers (remove the objects) and then set the pointers to 0...
but never ever call a destructor !!!

[agamemnus]

The system calls all my destructors

ehm, you must not call a destructor on your own !!!
they'll be executed automatically when the instance gets deleted !!!
if you want to clear something manually then best is to create a "cleanup" function and call this, and there delete all the pointers (remove the objects) and then set the pointers to 0...
but never ever call a destructor !!!

I think you misunderstand...

I'm talking about my own destructors, not any Irrlicht destructors.

[Lonesome Ducky]

No he understands. I think he was under the impression you were actually calling the destructor, rather than deleting the pointer and having the destructor called automatically.

[Acki]

I'm talking about my own destructors

hmm, maybe you shouldn't call it "destructor" but "cleanup function" !?!?!
constructor/destructor are fixed terms in C++...

a good idea to prevent missunderstandings is to show the code !!!
in this case how your cleanup function and destructor looks like and how/when you call the cleanup...

[agamemnus]

It's in Freebasic... doubtful anyone here knows it.

I avoided the problem by simply not calling destructors at the end of the program. I'll let the OS handle that.

[terier]

I'll let the OS handle that.

You shouldn't rely too much on your OS. Instead, try to understand how destructors work and how to properly delete object instances.