[PATCH v2] libstdc++: Use hidden friends for __normal_iterator operators
Jonathan Wakely
jwakely@redhat.com
Thu Jun 12 08:22:18 GMT 2025
On Thu, 12 Jun 2025 at 09:21, Jonathan Wakely <jwakely@redhat.com> wrote:
>
> On Thu, 12 Jun 2025 at 08:12, Daniel Krügler <daniel.kruegler@gmail.com> wrote:
> >
> > Am Do., 12. Juni 2025 um 00:10 Uhr schrieb Jonathan Wakely <jwakely@redhat.com>:
> >>
> >> As suggested by Jason, this makes all __normal_iterator operators into
> >> friends so they can be found by ADL and don't need to be separately
> >> exported in module std.
> >>
> >> The operator<=> comparing two iterators of the same type is removed
> >> entirely, instead of being made a hidden friend. That overload was added
> >> by r12-5882-g2c7fb16b5283cf to deal with unconstrained operator
> >> overloads found by ADL, as defined in the testsuite_greedy_ops.h header.
> >> We don't actually test that case as there's no unconstrained <=> in that
> >> header, and it doesn't seem reasonable for anybody to define such an
> >> operator<=> in C++20 when they should constrain their overloads properly
> >> (e.g. using a requires-clause). The heterogeneous operator<=> overloads
> >> added for reverse_iterator and move_iterator could also be removed, but
> >> that's not part of this commit.
> >>
> >> I also had to reorder the __attribute__((always_inline)) and
> >> [[nodiscard]] attributes, which have to be in a particular order when
> >> used on friend functions.
> >>
> >> libstdc++-v3/ChangeLog:
> >>
> >> * include/bits/stl_iterator.h (__normal_iterator): Make all
> >> non-member operators hidden friends, except ...
> >> (operator<=>(__normal_iterator<I,C>, __normal_iterator<I,C>)):
> >> Remove.
> >> * src/c++11/string-inst.cc: Remove explicit instantiations of
> >> operators that are no longer templates.
> >> * src/c++23/std.cc.in (__gnu_cxx): Do not export operators for
> >> __normal_iterator.
> >> ---
> >>
> >> v2: removed the unnecessary operator<=>, removed std.cc exports, fixed
> >> other minor issues noticed by Patrick.
> >>
> >> Tested x86_64-linux.
> >>
> >> libstdc++-v3/include/bits/stl_iterator.h | 327 ++++++++++++-----------
> >> libstdc++-v3/src/c++11/string-inst.cc | 11 -
> >> libstdc++-v3/src/c++23/std.cc.in | 9 -
> >> 3 files changed, 169 insertions(+), 178 deletions(-)
> >>
> >> diff --git a/libstdc++-v3/include/bits/stl_iterator.h b/libstdc++-v3/include/bits/stl_iterator.h
> >> index 478a98fe8a4f..a7188f46f6db 100644
> >> --- a/libstdc++-v3/include/bits/stl_iterator.h
> >> +++ b/libstdc++-v3/include/bits/stl_iterator.h
> >> @@ -1164,188 +1164,199 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> >> const _Iterator&
> >> base() const _GLIBCXX_NOEXCEPT
> >> { return _M_current; }
> >> - };
> >>
> >> - // Note: In what follows, the left- and right-hand-side iterators are
> >> - // allowed to vary in types (conceptually in cv-qualification) so that
> >> - // comparison between cv-qualified and non-cv-qualified iterators be
> >> - // valid. However, the greedy and unfriendly operators in std::rel_ops
> >> - // will make overload resolution ambiguous (when in scope) if we don't
> >> - // provide overloads whose operands are of the same type. Can someone
> >> - // remind me what generic programming is about? -- Gaby
> >> + private:
> >> + // Note: In what follows, the left- and right-hand-side iterators are
> >> + // allowed to vary in types (conceptually in cv-qualification) so that
> >> + // comparison between cv-qualified and non-cv-qualified iterators be
> >> + // valid. However, the greedy and unfriendly operators in std::rel_ops
> >> + // will make overload resolution ambiguous (when in scope) if we don't
> >> + // provide overloads whose operands are of the same type. Can someone
> >> + // remind me what generic programming is about? -- Gaby
> >>
> >> #ifdef __cpp_lib_three_way_comparison
> >> - template<typename _IteratorL, typename _IteratorR, typename _Container>
> >> - [[nodiscard, __gnu__::__always_inline__]]
> >> - constexpr bool
> >> - operator==(const __normal_iterator<_IteratorL, _Container>& __lhs,
> >> - const __normal_iterator<_IteratorR, _Container>& __rhs)
> >> - noexcept(noexcept(__lhs.base() == __rhs.base()))
> >> - requires requires {
> >> - { __lhs.base() == __rhs.base() } -> std::convertible_to<bool>;
> >> - }
> >> - { return __lhs.base() == __rhs.base(); }
> >> + template<typename _Iter>
> >> + [[nodiscard, __gnu__::__always_inline__]]
> >> + friend
> >> + constexpr bool
> >> + operator==(const __normal_iterator& __lhs,
> >> + const __normal_iterator<_Iter, _Container>& __rhs)
> >> + noexcept(noexcept(__lhs.base() == __rhs.base()))
> >> + requires requires {
> >> + { __lhs.base() == __rhs.base() } -> std::convertible_to<bool>;
> >> + }
> >> + { return __lhs.base() == __rhs.base(); }
> >>
> >> - template<typename _IteratorL, typename _IteratorR, typename _Container>
> >> - [[nodiscard, __gnu__::__always_inline__]]
> >> - constexpr std::__detail::__synth3way_t<_IteratorR, _IteratorL>
> >> - operator<=>(const __normal_iterator<_IteratorL, _Container>& __lhs,
> >> - const __normal_iterator<_IteratorR, _Container>& __rhs)
> >> - noexcept(noexcept(std::__detail::__synth3way(__lhs.base(), __rhs.base())))
> >> - { return std::__detail::__synth3way(__lhs.base(), __rhs.base()); }
> >> + [[nodiscard, __gnu__::__always_inline__]]
> >> + friend
> >> + constexpr bool
> >> + operator==(const __normal_iterator& __lhs, const __normal_iterator& __rhs)
> >> + noexcept(noexcept(__lhs.base() == __rhs.base()))
> >> + requires requires {
> >> + { __lhs.base() == __rhs.base() } -> std::convertible_to<bool>;
> >> + }
> >> + { return __lhs.base() == __rhs.base(); }
> >>
> >> - template<typename _Iterator, typename _Container>
> >> - [[nodiscard, __gnu__::__always_inline__]]
> >> - constexpr bool
> >> - operator==(const __normal_iterator<_Iterator, _Container>& __lhs,
> >> - const __normal_iterator<_Iterator, _Container>& __rhs)
> >> - noexcept(noexcept(__lhs.base() == __rhs.base()))
> >> - requires requires {
> >> - { __lhs.base() == __rhs.base() } -> std::convertible_to<bool>;
> >> - }
> >> - { return __lhs.base() == __rhs.base(); }
> >> -
> >> - template<typename _Iterator, typename _Container>
> >> - [[nodiscard, __gnu__::__always_inline__]]
> >> - constexpr std::__detail::__synth3way_t<_Iterator>
> >> - operator<=>(const __normal_iterator<_Iterator, _Container>& __lhs,
> >> - const __normal_iterator<_Iterator, _Container>& __rhs)
> >> - noexcept(noexcept(std::__detail::__synth3way(__lhs.base(), __rhs.base())))
> >> - { return std::__detail::__synth3way(__lhs.base(), __rhs.base()); }
> >> + template<typename _Iter>
> >> + [[nodiscard, __gnu__::__always_inline__]]
> >> + friend
> >> + constexpr std::__detail::__synth3way_t<_Iterator, _Iter>
> >> + operator<=>(const __normal_iterator& __lhs,
> >> + const __normal_iterator<_Iter, _Container>& __rhs)
> >> + noexcept(noexcept(std::__detail::__synth3way(__lhs.base(), __rhs.base())))
> >> + requires requires {
> >> + std::__detail::__synth3way(__lhs.base(), __rhs.base());
> >> + }
> >> + { return std::__detail::__synth3way(__lhs.base(), __rhs.base()); }
> >> #else
> >> - // Forward iterator requirements
> >> - template<typename _IteratorL, typename _IteratorR, typename _Container>
> >> - _GLIBCXX_NODISCARD __attribute__((__always_inline__)) _GLIBCXX_CONSTEXPR
> >> - inline bool
> >> - operator==(const __normal_iterator<_IteratorL, _Container>& __lhs,
> >> - const __normal_iterator<_IteratorR, _Container>& __rhs)
> >> - _GLIBCXX_NOEXCEPT
> >> - { return __lhs.base() == __rhs.base(); }
> >> + // Forward iterator requirements
> >> + template<typename _Iter>
> >> + __attribute__((__always_inline__)) _GLIBCXX_NODISCARD
> >
> >
> > Just curious: So in this case the attribute order intentionally potentially deviates from the order of the other friend functions using [[nodiscard, __gnu__::__always_inline__]]?
>
> Yes because operator== needs to compile as C++98 so it uses
> __attribute__((always_inline)) not [[gnu::always_inline]], and Clang
> will only allow __attribute__((foo)) [[bar]] and not the other way
> around:
> https://godbolt.org/z/nGWss9YG3
>
> <source>:5:36: error: an attribute list cannot appear here
> 5 | __attribute__((always_inline)) [[nodiscard]]
> | ^~~~~~~~~~~~~
I'll update the commit msg to note that the order is needed for Clang,
as GCC seems happy with either order.
>
> We could change [[nodiscard,gnu::always_inline]] to be in the opposite
> order, but when either order works, I prefer to have the standard
> attribute first and the non-standard one second.
More information about the Libstdc++
mailing list