https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92894
Patrick Palka <ppalka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |ppalka at gcc dot gnu.org
--- Comment #3 from Patrick Palka <ppalka at gcc dot gnu.org> ---
(In reply to Jonathan Wakely from comment #0)
> The following should compile, because the undefined member function is never
> odr-used, only its return type is needed.
Hmm.. don't we end up odr-using the undefined member function when checking the
{ ranges::iter_move(__in) } -> same_as<iter_rvalue_reference_t<_In>>;
requirement of the concept __indirectly_readable_impl (as part of the concept
indirect_unary_predicate)?
In order to verify this requirement we need to determine the return type of
ranges::iter_move(__in). But the operator() of this CPO has a decltype(auto)
return type, so determining its return type requires instantiating its body:
struct _IMove
{
template<typename _Tp>
requires __adl_imove<_Tp> || requires(_Tp& __e) { *__e; }
constexpr _Blah<_Tp>::type
operator()(_Tp&& __e) const
noexcept(_S_noexcept<_Tp>())
{
if constexpr (__adl_imove<_Tp>)
return iter_move(static_cast<_Tp&&>(__e));
else if constexpr (is_reference_v<iter_reference_t<_Tp>>)
return std::move(*__e);
else
return *__e;
}
};
And this instantiated function body (the second branch of the constexpr if)
would then odr-use this undefined member function, IIUC.
So could this be another example where defining the operator() of a CPO with a
deduced return type leads to excessive instantiation?
