As predicted, size_t and ssize_t is back to normal.  (Hasso already noted this in a reply to the original post.)