On Fri, 27 Jan 2023 at 14:30, Xi Ruoyao wrote:
>
> On Fri, 2023-01-27 at 14:07 +0000, Jonathan Wakely via Gcc-help wrote:
> > I think it's safe to assume that *either* the code compiles and works
> > as expected, or fails to compile. And in practice it compiles and
> > works with all widely used compilers. You will not find a C++
> > implementation where the types are not compatible, but the code
> > compiles anyway and then misbehaves at runtime.
>
> FWIW if we do "some strange thing" we may end up shooting our own foot.
> Like:
>
> a.cc:
>
> #include <cstdio>
>
> extern "C"
> {
> struct A { struct {} x; int a; };
> void callback(void (*fn)(struct A *));
> }
>
> void f(struct A *p)
> {
> std::printf("%d\n", p->a);
> }
>
> int main()
> {
> callback(f);
> }
>
> b.c:
>
> #include <stdio.h>
>
> struct A { struct {} x; int a; };
> void callback(void (*fn)(struct A *))
> {
> struct A foo = { .a = 42 };
> fn(&foo);
> }
>
> This thing won't work with GCC because struct A will have different
> layouts in GNU C and C++. Note that the C standard does not allow an
> empty struct at all (a pedantic C compiler should reject b.c as an empty
> struct violates the syntax rule of C). But GNU C supports it as an
> extension.
Strictly speaking, a pedantic C compiler should issue a diagnostic. It
doesn't have to fail to compile.
$ gcc -pedantic b.c
b.c:3:12: warning: struct has no members [-Wpedantic]
3 | struct A { struct {} x; int a; };
| ^~~~~~
