Re: dealloc (Correction)

Lim Hock-Chai wrote:

The compiler checking I'm referring to is more like in the source code
you do something like this: myField_BasePointer = *null;  (from
everybody's posting, seems like you should always do dealloc(n)
myField_BasePoniter)

No, it's the other way around.  If you dealloc myField_BasePointer, you
should make sure you null that variable, and any other variable that has
a pointer into that heap storage.

This is similar to trying to relate deleting a file to clearing the
file's name (I think I'm stealing from Scott K here).  You are
suggesting the rule: "If you clear a variable that holds a file's name,
then you should delete the file".  A more reasonable rule would be: "If
you delete a file, then you should clear any variables that hold the
file's name".

I can't remember which of these points has already been made in the
various versions of this thread.  Sorry for any repetitions.
 - A basing pointer isn't necessarily a pointer to heap storage.  You
could set 
   the basing pointer to %ADDR of some other field.
 - The address might be saved in another pointer and still be needed by
the application
 - The DEALLOC has to be done with the original address; if you advance
the pointer 
   within your allocated storage, the DEALLOC would not work.

     p = %alloc(100);
     for i = 1 to 10;
       p = p + 1;
     endfor;
     dealloc p;  <---- this won't work

   The dealloc will not deallocate the storage because p now points 10 
   bytes past the original starting address of the allocation.