|
So its dangerous to do something like I did in that example as p may access a piece of memory that does not exist?
|
|
|
|
|
minkowski wrote: So its dangerous to do something like I did in that example
Yes it is. The behaviour is undefined.
minkowski wrote: as p may access a piece of memory that does not exist?
The memory do still exist. However, the content may have change.
Regards,
Tim
|
|
|
|
|
In addition to what already been said, I would present the safe way of doing it: declare the variables as static, like this:
<code>
int* create( )
{
static int marks = 5;
static int *pt=&marks;
return pt;
}
</code>
If you declare variables static, they will be stored in the heap instead of the stack.
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
That could be equally dangerous, if you ever plan on calling the create function multiple times
|
|
|
|
|
Correct me if I'm wrong, but as far as I remember, a static variable gets a fixed position on the heap.
If I'm right, what's the danger with calling the function more than once?
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
int *p;
int *q;
p=create();
q=create();
(*q)++;
cout<<endl<<"The marks are : "<<*(p);
return 0;
p and q will both point to the same location, so modifying one will modify the other, which might not be the expected behaviour
|
|
|
|
|
here is an example of where your code would not work :
::printf("%d %d", *(create()), *(create()));
|
|
|
|
|
Correct me if I'm wrong, but as far as I remember, a static variable gets a fixed position on the heap.
If I'm right, what's the danger with calling the function more than once?
-- modified at 7:59 Wednesday 25th October, 2006
Now I've checked your code. It returns 5 both times.
BTW, you mean
::printf("%d %d", *(create()), *(create()));
don't you?
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
kakan wrote: what's the danger with calling the function more than once?
if the functions does this for instance :
int* create() {
static int p = 0;
p++;
return &p;
}
with such printf(), all the calls will print the same value...
kakan wrote: BTW, you mean
::printf("%d %d", *(create()), *(create()));
yes ^^ Fixed, thanks 
|
|
|
|
|
Nope, a static gets inititalized just once.
So with your printf, the result gets this:
2 1
Well. I don't want to be rude, but if you change a variable, then the variable in question obviously gets changed...
static or not.
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
no, do the test... the result will be 2 2
it is initialized once, but after that, the value it gets incremented is the same at final...
|
|
|
|
|
Funny, in my VC6, this code:
<br />
int* create( )<br />
{<br />
static int p = 0;<br />
p++; <br />
return &p;<br />
}<br />
<br />
main()<br />
{<br />
::printf("%d %d", *(create()), *(create()));<br />
}<br />
Produces the result:
2 1
If I set a breakpoint at "return &p", p contains 1 for the first call, and 2 for the second call.
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
humm, what about when the function returns p directly, instead of its address ?
i know i already encountered such a problem, but i don't remember the exact case...
|
|
|
|
|
Same result, this code:
<br />
int create2( )<br />
{<br />
static int p = 0;<br />
p++; <br />
return p;<br />
}<br />
<br />
::printf("%d %d", create2(), create2());<br />
produces the result:
2 1
Same as before. A breakpoint at "return p" gives the same result as before (of course).
Have you tried it yourself?
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
Hi kakan
isnt that what you would expect because the definition of a static variable is to create 1 and only 1 version of the variable?
|
|
|
|
|
Hello. That's exactly what I expect. But tox obviously doesn't. And it seems as it's my job to prove him wrong, instead of the opposite...
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
Sorry, may be I am getting confused....  Um, since at every funcion call to create2() you set p = 0 it should reset p everytime. But appears to not. Why is that?
Thanks for any information.
|
|
|
|
|
Sorry, may be I am getting confused In your definition of create2() you reset p, by p = 0. So in each call to create2() p should be set to 0 before being incremented. So wouldnt you therefore expect the same value printed everytime?
|
|
|
|
|
Sorry, may be I am getting confused In your example above of create() you reset p, by p = 0. So in each call to create() p should be set to 0 before being incremented. So wouldnt you therefore expect the same value printed everytime?
|
|
|
|
|
Actually I just answered my own question
The static int p = 0 only gets called once at instantiation which makes sense coz otherwise you will have loads of p's around which goes against the idea of it being static (only 1 copy of the variable exists).
Well I hope that is correct.
|
|
|
|
|
Sorry for the delay. It's due to the fact I went home from work yesterday. 
That's right, a static variable just gets initialized once.
Besides, there are just one copy of it, in the heap.
The fact that it is stored in the heap (as opposite to the stack), makes it protected from alterations.
So if you want to make sure (as in this case) that a value, returned from a function will contain the same value, static is the way to go.
This goes for plain C. For C++, a class member variable does the same.
Kakan
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
Hello
Are you sure that static members are on the heap? That would imply they are dynamically allocated would'nt it? Also if you have an item on the heap you would need to use 'delete' at some point for clean up. If you can send me a reference for this I would be most interested.
Thanks !
|
|
|
|
|
Sorry, I was wrong about this, it seems.
The most common place to store static variabels seems to be in the data segment of the program.
Sorry. 
But the rest is true! 
Alcohol. The cause of, and the solution to, all of life's problems - Homer Simpson
|
|
|
|
|
I did a google on 'static heap C++ allocation'. It seems on some systems its on the heap and others on the stack. But it is allocated at run time.
|
|
|
|
|
kakan wrote: Correct me if I'm wrong, but as far as I remember, a static variable gets a fixed position on the heap.
Nope. Not on the heap.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
