[expr.dynamic.cast] CWG 2365: Incorrect wording regarding dynamic_cast

[introom]

In paragraph 4, http://eel.is/c++draft/expr.dynamic.cast#4

If the value of v is a null pointer value in the pointer case, the result is the null pointer value of type T.

And in paragraph6,

Otherwise, v shall be a pointer to or a glvalue of a polymorphic type.

So the otherwise means the v in p4 doesn't have to be polymorphic.

However, this is not possible because the compiler can only check if "v" is a null pointer at runtime. If "v" is not a null pointer, and unluckily has no RTTI, then the runtime could do nothing meaningufl, ill-formed?

BTW, Clang and GCC requires the "v" in p4 to be of pointer to polymorphic type.

[cpplearner]

Do you mean this:

struct T1 {};
struct T2 {};
T1* p = nullptr;
dynamic_cast<T2*>(p);

This appears to be valid according to p4, but will not be valid if p is not a null pointer.

[jensmaurer]

This does not seem to be editorial; the normative wording is nonsensical here. Submitted as a CWG issue: http://lists.isocpp.org/core/2017/02/1733.php .

[jensmaurer]

CWG in Jacksonville: treat as a core issue

[jensmaurer]

Resolved with commit e81049e.