Hi
I have been following the apr memory thread with great interest and I
think I am missing some point that maybe all of you folks know. I am
referring to the concept of using shared memory and regular memory as
a "one" pool resource.
Consider the following example...
struct foo {
void *ptr;
};
struct foo *myFoo;
myFoo = (struct foo *)apr_shm_malloc( ..);
tmpStruct->ptr = apr_palloc( ...);
or the following could be coded
tmpStruct->ptr = apr_shm_malloc( ...);
The problem here is that you putting both regular memory and shared
memory under one hat. As a result a programmer could have the
impression that it is regular memory that can be referenced using a
simple pointer. But the truth is that allocating shared memory and
then using pointers within shared memory is an absolute pain in the
butt. OR one could use the VC++ compiler with its support of offset
based pointers. And I suppose everyone is ready to jump to VC++ ;)
Seriously though, how would the use of pointers in structures work
when using shared memory and regular memory allocated from "one" pool
work?
Christian Gross
