In C++ Templates - The complete guide, 2nd edition, at page 434, a macro is defined to generate predicates testing the existence of a nontype memeber in a class:
#include <type_traits>
#define DEFINE_HAS_MEMBER(Member) \
template<typename T, typename = void> \
struct HasMemberT_##Member : std::false_type {}; \
template<typename T> \
struct HasMemberT_##Member<T, std::void_t<decltype(&T::Member)>> : std::true_type {};
The the text reads
It would not be difficult to modify the partial specialization to exclude cases wehre
&T::Member
is not a pointer-to-member type (which amounts to excludingstatic
data members).
So I've played around with static_assert
s and compiler errors and remember about types of static
and non-static
of members of classes:
struct A {
int begin;
static int end;
};
static_assert(std::is_same<decltype(&A::begin), decltype(A::begin) A::*>::value, ""); // passes
static_assert(std::is_same<decltype(&A::end), decltype(A::end)*>::value, ""); // passes
And thought that maybe a good modification of the lambda above aimed at detecting non-static
members only is to change
: std::true_type
to
: std::is_same<decltype(&T::Member), decltype(T::Member) T::*>
whereas if I wanted to check for static
memebers only, I could change it to
: std::is_same<decltype(&T::Member), decltype(T::Member)*>
Is it really this easy? Or am I overlooking something important?
You are overlooking non-static member functions.
The primary use case of the traits created by DEFINE_HAS_MEMBER
is to check for the existence of a particular method. To borrow from the terminology in the question, a range is defined by having begin()
and end()
. While the trait created by the original macro works, your extra condition does not.
prog.cc:7:61: error: invalid use of non-static member function 'int A::begin()'
7 | static_assert(std::is_same<decltype(&A::begin), decltype(A::begin) A::*>::value, "");
| ^~~~~
prog.cc:7:72: error: template argument 2 is invalid
7 | static_assert(std::is_same<decltype(&A::begin), decltype(A::begin) A::*>::value, "");
|
The reason in standartese is this:
[expr.prim.id]
2 An id-expression that denotes a non-static data member or non-static member function of a class can only be used:
- as part of a class member access in which the object expression refers to the member's class or a class derived from that class, or
- to form a pointer to member ([expr.unary.op]), or
- if that id-expression denotes a non-static data member and it appears in an unevaluated operand. [ Example:
— end example ]struct S { int m; }; int i = sizeof(S::m); // OK int j = sizeof(S::m + 42); // OK
These are the only valid uses of something like T::Member
. Non static member functions don't hit any of the bullets, so the construct is ill-formed when used to define the body of HasMemberT_##Member
's specialization. No SFINAE, just a hard error.