CWG2531 [dcl.constexpr] Static data members redeclared as constexpr

[opensdh]

Reference (section label)

[dcl.constexpr]/1

Issue description

C++17 made constexpr static data members implicitly inline:

A function or static data member declared with the constexpr or consteval specifier is implicitly an inline function or variable ([dcl.inline]).

However, that makes the following well-formed C++14 program IFNDR per [dcl.inline]/5:

// x.hh
struct X {
  static const int x;
};

// TU 1
#include"x.hh"
constexpr int X::x{};

// TU 2
#include"x.hh"
int main() {return !&X::x;}

Current GCC and Clang do treat X::x as a weak symbol, but since they emit it even without an odr-use in the TU programs like this do work in practice.

Suggested resolution

Change [dcl.constexpr]/1 as follows:

[…] A function or static data member declared with the constexpr or consteval specifier on its first declaration is implicitly an inline function or variable ([dcl.inline]). […]

(Functions must be declared constexpr on every declaration if on any, so this isn't a change for them.)

[jensmaurer]

Reflector thread: http://lists.isocpp.org/core/2022/02/12017.php

[jensmaurer]

@opensdh, please check: CWG2531

[opensdh]

It looks correct; presumably you omitted the implementation behavior purposefully.

[jensmaurer]

Yeah, I'm not sure it adds much, given that we want to outlaw the situation regardless.