[Bug c/67882] surprising offsetof result on an invalid array member without diagnostic

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67882

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|SUSPENDED                   |RESOLVED
         Resolution|---                         |FIXED

--- Comment #10 from Martin Sebor <msebor at gcc dot gnu.org> ---
The sentiment within WG14 about the question raised in comment #5 was that the
use case is invalid and the offsetof description makes it sufficiently clear. 
In

  struct s {
    int fpr[32][1];
    int fpscr;
  };

  offsetof(struct s, fpr[32][0])

the offsetof macro is specified to expand such that given

  static struct s t;

the following evaluates to an integer constant:

  &(t.fpr[32][0])

Since the t.fpr array has only 32 elements, a reference to t.fpr[32][0] is out
of bounds and thus undefined (there is no array fpr[32] and so no element 0 of
such an array).  As a result, the full expression does not yield a valid
integer constant as required and a diagnostic is justified.  With that I think
it's time to resolve this.

(Other questions were raised about offsetof in DR 496 but even those the
committee doesn't seem to be very interested in dealing with.)