https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115909
Bug ID: 115909
Summary: C++20 operator<=> explicitly defaulted but defined as
deleted after first declaration does not error
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: mital at mitalashok dot co.uk
Target Milestone: ---
When a function is defined as deleted (even if it was explicitly defaulted), it
must be the first declaration
<https://eel.is/c++draft/dcl.fct.def.delete#4.sentence-3>.
However, GCC doesn't appear to issue an error when this happens for a
operator<=>.
https://godbolt.org/z/KseW3r8EW
```
namespace std {
struct strong_ordering { static const strong_ordering less, equal, greater;
};
}
struct X {
X(int);
};
struct Y {
X x;
Y();
friend auto operator<=>(Y, Y);
};
struct Z {
X x;
auto operator<=>(const Z&) const;
};
// Y::Y() = default;
auto operator<=>(Y, Y) = default;
auto Z::operator<=>(const Z&) const = default;
```
Y::Y() is rejected as expected when uncommented (although the error could use a
little work? It should say something like "error: deleted definition must be
the first declaration"), but with the comment this compiles without any
diagnostics.
If you replace the `= default` with `= delete`, this compiles with a warning
only when it should error (Which is Bug 52659 I think)
Possibly related: Bug 103947
