http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47939
--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-03-01
14:31:35 UTC ---
The lack of knowledge of George_t is the worst side-effect of this bug.
/* { dg-do run } */
/* { dg-options "-g" } */
typedef struct _Harry { int dummy; } Harry_t;
Harry_t harry; /* { dg-final { gdb-test 12 "((Harry_t *)&harry)->dummy" "0" } }
*/
typedef const struct _George { int dummy; } George_t;
George_t george; /* { dg-final { gdb-test 12 "((George_t *)&george)->dummy" "0"
} } */
typedef const Harry_t Michael_t;
Michael_t michael; /* { dg-final { gdb-test 12 "((Michael_t *)&michael)->dummy"
"0" } } */
int main()
{
return 0;
}
currently George_t and Michael_t using tests are UNSUPPORTED (error in
processing the command).
The patch has weird side-effects on diagnostics, it only shows a possible
hint as to what the reason is we drop the TYPE_NAME.
