https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128603
Bug ID: 128603
Summary: __is_array is inconsistent with overload resolution on
zero size arrays
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: nikolasklauser at berlin dot de
Target Milestone: ---
Consider this beautiful piece of code:
```
struct S {
int a[0];
};
static_assert(__is_array(S[]));
static_assert(__is_array(S[10]));
template <class T, int N>
constexpr int func(T(&)[N]) {
return N;
}
void test() {
S arr[10];
static_assert(func(arr) == 10);
}
```
The first `static_assert` passes, while the second fails. The third passes
again. I don't know why GCC behaves this way, but it seems inconsistent. Clang
accepts this.
P.S. I'd be perfectly happy if the conclusion here is that arrays of zero size
types should be rejected. I don't see a scenario where this is useful other
than to file bugs against compilers (I've found this through a libc++ test
case).
