Suggestions and question about bool pointers

Made a new thread 'cause I suspected no one would read the other one, 'cause I said it was fixed. Don't hate me, please. xP

http://codepad.org/opw4HeWp

Now, I'd like suggestions on this and is there anyway to check if a pointer, like the used pFrameFunc pointer, has been assigned anything else than 'garbage' ? Like, if it has been assigned by the code, and not some default compiler thing...

And for the suggestions, I mean would this be the right place for the loop? I don't want the users of this engine to make the loop themselves...
And I use the variable "bool& failed" as the condition for the loop, there is no problem with this? I mean, it is a pointer... Just to make me more sure about this!
Thanks in advance!

What you have posted makes no sense with respect to the question, since you deleted pFrameFunc and any reference to it, et cetera.

Since it was (as I recall) passed in to your function, then it would have to have come from somewhere. If you want to make sure that it always has a valid value, then you want to set it to something in the constructor, presumably.

I'm pretty sure you never want a constructor to loop infinitely.

And no, your failed variable is not a pointer.

well, this constructor will be called like LIQUID* liq; then later when all is set to go, liq = new LIQUID(kinda many arguments); personally I see no flaw in this, but I've been wrong plenty before ;)

Code:

---


LIQUID* liq;
SYSTEM sys;

int main(){

    sys.SetFrameFunc(SOME_DEFINED_FUNCTION());

    bool initCheck;

    liq = new LIQUID(initCheck, sys, ect...);
    if(initCheck) return true;

    return 0;
}

---

Just a short and fast example of how it will be called(most likely)

Quote:

---

Originally Posted by Akkernight

well, this constructor will be called like LIQUID* liq; then later when all is set to go, liq = new LIQUID(kinda many arguments); personally I see no flaw in this, but I've been wrong plenty before ;)

---

This constructor won't be called when you do LIQUID *liq, since you aren't creating a LIQUID object at that time. When you do call the constructor, you don't get kinda many arguments. You get two arguments, since that's what your constructor takes. Notice that that means, in particular, that you will never get to set up your callback function, since that's not one of your arguments, and you will never get a chance to call your set-up function, since no more lines of your main program will ever execute (since you said that this function should run until shutdown).

If I understand you correctly, I understand that you think I'm a complete retard...

The engine isn't done yet, infact far from. What that sketch showed, was how it WILL be.
Also, I've changed too much of the design for all before this post to be really 'valid' to consider once you post, or... I dunno how to explain, but the code is changed.

But I do have one question... If I call " return pFrameFunc; " does it run through pFrameFunc? Meaning, it runs the function that pFrameFunc points to ?

Quote:

---

Originally Posted by Akkernight

The engine isn't done yet, infact far from. What that sketch showed, was how it WILL be.
Also, I've changed too much of the design for all before this post to be really 'valid' to consider once you post, or... I dunno how to explain, but the code is changed.

---

Well, fair enough, but then asking questions about the code that isn't there isn't going to get you very far.

Quote:

---

But I do have one question... If I call " return pFrameFunc; " does it run through pFrameFunc? Meaning, it runs the function that pFrameFunc points to ?

---

If you do return 5; does C++ try to run the function at memory address 5? Or does it just return the value 5?

that's a number... Won't return need to run the function to check or something? Or do I need to make a RunFrameFunc() function too ?

EDIT: And when I posted this, I didn't have a plan to change it or anything, but I did change it after your reply, it was just so that all future posts think of that.. Well, that kinda changes the question of the topic :S

Quote:

---

Originally Posted by Akkernight

that's a number...

---

So is pFrameFunc. What's your point?

Quote:

---

Originally Posted by Akkernight

Won't return need to run the function to check or something? Or do I need to make a RunFrameFunc() function too ?

---

Return doesn't "need to" do anything. If you want the function to be called, then call it. There's no need to make yet another function to pass the function pointer to and so on infinitely; just stick parentheses at the end of it like any other function.

... I want to run the pointer, like, call pFrameFunc somewhere, and it runs the function which pFrameFunc points to, also the reason I asked if I was to make a new function is 'cause I amde a new class called SYSTEM which has this stuff now, and the pointers are private....

If you have access to the pointer, then you can run it. If pFrameFunc is now a private member of sys, then you're going to have to run it through that object, yes. Of course, in this case it wouldn't be possible to even consider "return pFrameFunc;" since pFrameFunc would not be available to us to use (being a member of some other class).

ok, thanks!

>>sys.SetFrameFunc(SOME_DEFINED_FUNCTION());
Note that this will run SOME_DEFINED_FUNCTION and pass what is returned from that function to SetFrameFunc.
If you want to assign a function pointer, then take its address instead of calling it:
sys.SetFrameFunc(&SOME_DEFINED_FUNCTION);

2nd) I don't think it's a good idea the for constructor to block. Not at all. The constructor should only initialize the object, making it ready for use. Create some other function, such as Run, which can block.

3rd) I believe it's a better idea to throw an exception is the constructor fails rather than return a failure. Your approach just says it "failed", as well, but not why, which is not very helpful.

First I make it work, then I make it awesome ;)

That said, though, it's usually not too helpful to go the "wrong way" too far since it will end up costing more time to make the "correct" changes.
Like throwing an exception - just a simple one can't be too much work for the basic code.

Quote:

---

Originally Posted by Akkernight

Now, I'd like suggestions on this and is there anyway to check if a pointer has been assigned anything else than 'garbage' ? Like, if it has been assigned by the code, and not some default compiler thing...

---

No. There is not a way to do that.
Given an arbitrary value for a pointer, you can certainly check if it happens to be a value close to NULL, or equal to a compiler filled value like 0xCCCCCCCC, and if you really get your hands dirty, you can with a high degreee of accuracy tell if is points to within the stack vs the heap. However, none of this will EVER guarantee you that it points to something valid.

You must never write a program such that you have to determine afterwards whether a pointer value is valid or not. You must always already know whether it is valid or not, and this often means tracking it yourself. Commonly we track an invalid pointer by setting it to NULL. Other ways might be to have a decicated boolean variable alongside the pointer.
If you ignore this advice you're going to screw yourself.