https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125663
Bug ID: 125663
Summary: [reflection] Reflecting a conversion-function-id of a
conversion function template yields a reflection of
the template, but it should yield the named
specialization
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: ensadc at mailnesia dot com
Target Milestone: ---
https://gcc.godbolt.org/z/sMbE3zo7W
====
#include <meta>
struct A {
template<class T>
operator T();
};
constexpr auto r = ^^A::operator int;
static_assert(std::meta::is_function(r));
====
<source>:9:37: error: static assertion failed
9 | static_assert(std::meta::is_function(r));
| ~~~~~~~~~~~~~~~~~~~~~~^~~
====
It appears that GCC finds the name `operator int` in `^^A::operator int` to
correspond to a function template, and yields a reflection of that template.
This behavior makes sense if the name were an identifier, but here the name is
a conversion-function-id.
Since a conversion-function-id is not a *reflection-name* and is a
*id-expression*, the *reflect-expression* should be subject to
[expr.reflect]/7, which says that:
> if the id-expression denotes an overload set S, overload resolution for the
> expression &S with no target shall select a unique function ([over.over]);
> R represents that function.
That is, the result of `^^A::operator int` should represent a function, not a
function template.
