Inline...
On Sun, 2004-05-23 at 10:56, Simon Ask Ulsnes wrote:
<snip/>
> >Furthermore, if you do something like this:
> >
> > const char*
> > get_my_string()
> > {
> > std::string s ("this is my string");
> > return s.c_str();
> > }
> >
> >You're *asking* for trouble, as the std::string destructor will free the
> >memory used to hold the string, so the string returned by
> >get_my_string() will be pointing to invalid memory.
>
> And guess what, that's actually what I'm doing (basically) - so far, it
> works, I haven't detected any memory leaks (haven't checked very
> thoroughly, though).
> But you say using marshalling and IntPtr's is the best way to do it?
The above situation isn't a memory leak, so it won't be detected as a
memory leak. It's instead a "use after free". It's akin to doing this:
char *mem = malloc (20); // allocate
strcpy(mem, "some string"); // initialize
free (mem); // free
printf ("this is my string: %s\n", mem); // use?! bad.
It's not safe to use memory after it's freed. Especially in a
multi-threaded environment -- the memory could have been re-allocated by
another thread and initialized with different data.
So this isn't a memory corruption bug or a memory leak, it's just bad
memory hygiene.
- Jon
_______________________________________________
Mono-list maillist - [EMAIL PROTECTED]
http://lists.ximian.com/mailman/listinfo/mono-list
