Update dynamic window example to use MPI_Aint_add

[mpiforumbot]

Originally by jdinan on 2013-12-03 10:38:25 -0600

---

__ This ticket is part of a series: #349, #402, #404 __*

-_ Note: This ticket is now stale. Please refer to the PDF formal proposal attached to #349. _**

Assuming that #349, which adds MPI_Aint_add, is passed. Update example 11.22, pg. 470 to remove casts from MPI_Aint to a native pointer.

Page 470, Line 15:

-an environment where the length of a pointer is different on different processes.
+an environment where the layout of the structures is different on different processes.

Page 470, Line 34:

+#define LLIST_ELEM_NEXT_RANK ( offsetof(llist_elem_t, next) + offsetof(llist_ptr_t, rank) )
+#define LLIST_ELEM_NEXT_DISP ( offsetof(llist_elem_t, next) + offsetof(llist_ptr_t, disp) )

Page 472, Line 10:

-   (MPI_Aint) &(((llist_elem_t*)tail_ptr.disp)->next.rank), llist_win);
+   MPI_Aint_add(tail_ptr.disp, LLIST_ELEM_NEXT_RANK), llist_win);

Page 472, Line 17:

-   (MPI_Aint) &(((llist_elem_t*)tail_ptr.disp)->next.disp), 1,
+   MPI_Aint_add(tail_ptr.disp, LLIST_ELEM_NEXT_DISP), 1,

Page 472, Line 29:

-   (MPI_Aint) &(((llist_elem_t*)tail_ptr.disp)->next.disp),
+   MPI_Aint_add(tail_ptr.disp, LLIST_ELEM_NEXT_DISP),

Change Log Entry

See #349.

[mpiforumbot]

Originally by goodell on 2014-02-05 15:40:12 -0600

---

Do these also need to be using MPI_Get_address in order to show a more portable example?

[mpiforumbot]

Originally by jsquyres on 2014-06-05 10:34:41 -0500

---

Ticket passed 1st vote: 16 yes, 0 no, 3 abstain

[mpiforumbot]

Originally by schulzm on 2014-12-10 17:51:59 -0600

---

Had second vote and passed at December 2014 meeting