[diff.expr] Unmentioned incompatibility with C about pointer indirection CWG2875

[frederick-vs-ja]

These usages of pointer indirection are explicitly allowed in C via WG14 N721, but seemly not allowed in C++.

int* p = &(*((int*)0)); // undefined behavior in C++, well-defined in C (DR076/N721)
int a[10];
int* q = &a[10];        // undefined behavior in C++, well-defined in C (DR076/N721)

Should we mention them in [diff.expr]?

[xmh0511]

This is CWG232. A good question located at https://stackoverflow.com/questions/47227977/is-dereferencing-invalid-pointers-legal-if-no-lvalue-to-rvalue-conversion-occurs/47229011#47229011

[frederick-vs-ja]

This is CWG232. A good question located at https://stackoverflow.com/questions/47227977/is-dereferencing-invalid-pointers-legal-if-no-lvalue-to-rvalue-conversion-occurs/47229011#47229011

Thank you!

---

I find that this change is essentially a DR in C (DR076). And the proposed resolution of CWG232 is significantly more permissive than that of C.

[frederick-vs-ja]

Reopening as CWG232 is closed as NAD.

Perhaps such incompatibility is intended.

[frederick-vs-ja]

Addressed by CWG2875.