[Bug c++/15942] modifying data in a int * cast pointer to a pointer to a member has no effect when -O2 is used

[indrek at mare dot ee]

------- Additional Comments From indrek at mare dot ee  2004-06-11 16:59 -------
I know it's a hack. But it's a hack that has consistently served me on varous
platforms (and compilers) considering c++ limitations at handling array of
member pointers. I got over it myself by using memcpy instead. Still c and c++
languages allow casts for some reason. Or if this is incorrect - then why does
gcc allow to compile it anyway.

Also this behaviour is not consistent. For example add printf("%x\n", &n[j]);
before the increment line, like this:

  for ( int j = 0; j < 8; j++ ) {
        n[j] = (int a::*)&a::x;
        printf("%x\n", &n[j]);
        *(int *)&n[j] += j*8;
  }

And it'll give you different (correct) output - the memory gets changed. Maybe
it's not a big deal for you but this might hint of other more sinister bug
inside gcc. I wont bother you any more.

Also this bug seems not to manifest on i386 platform or another arm compiler I used.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|INVALID                     |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15942