Alignment of foo[1][1][1][1]

[Harald van Dijk]

  int i;
  double *p;
  p = (double *) &i;
  assert((int *) p == &i);

However, that is the _only_ guarantee made about the value of p. Since
dereferencing p is not allowed in strictly conforming programs anyway,
the standard does not need to, and does not, address the question of
whether p points to any object at all.

[...]

I don't believe the standard guarantees that at all.

As before in this thread, a part of the sentence is missing. The point
was that the standard requires that the assert passes, if &i is
suitably aligned for a pointer of type double *. Sorry for the
confusion.

 

[Keith Thompson]

  int i;
  double *p;
  p = (double *) &i;
  assert((int *) p == &i);

However, that is the _only_ guarantee made about the value of p. Since
dereferencing p is not allowed in strictly conforming programs anyway,
the standard does not need to, and does not, address the question of
whether p points to any object at all.

[...]

I don't believe the standard guarantees that at all.

As before in this thread, a part of the sentence is missing. The point
was that the standard requires that the assert passes, if &i is
suitably aligned for a pointer of type double *. Sorry for the
confusion.

Hmm, I hadn't realized that round-trip pointer-to-pointer conversion is
guaranteed to yield the original value if the alignment is ok (C99
6.3.2.3p7). (I'm sure I had read that paragraph; I just forgot about
it.)