[PATCH] libstdc++: Fix std::format_kind primary template for Clang [PR120190]

[Tomasz Kaminski]

Please also add the message to dg-error check in format_kind_neg.cc.
With that LGTM.

On Thu, May 15, 2025 at 2:19 PM Jonathan Wakely <jwakely@redhat.com> wrote:

> On Thu, 15 May 2025 at 12:15, Daniel Krügler <daniel.kruegler@gmail.com>
> wrote:
> >
> > Am Do., 15. Mai 2025 um 13:00 Uhr schrieb Jonathan Wakely <
> jwakely@redhat.com>:
> >>
> >> Although Clang trunk has been adjusted to handle our std::format_kind
> >> definition (because they need to be able to compile the GCC 15.1.0
> >> release), it's probably better to not rely on something that they might
> >> start diagnosing again in future.
> >>
> >> Define the primary template in terms of an immediately invoked function
> >> expression, so that we can put a static_assert(false) in the body.
> >>
> >> The diagnostic that users will see looks like this:
> >>
> >> include/c++/16.0.0/format:5470:21: error: static assertion failed
> >>  5470 |       static_assert(false); // Cannot use primary template of
> std::format_kind
> >>       |                     ^~~~~
> >>
> >> libstdc++-v3/ChangeLog:
> >>
> >>         PR libstdc++/120190
> >>         * include/std/format (format_kind): Adjust primary template to
> >>         not depend on itself.
> >>         * testsuite/std/format/ranges/format_kind_neg.cc: Adjust
> >>         expected errors. Check more invalid specializations.
> >> ---
> >>
> >> Tested x86_64-linux.
> >>
> >> I haven't tested it fully with Clang, because the version of Clang in
> >> Fedora is older than the change which started rejecting the previous
> >> definition of std::format_kind, and the build in Compiler Explorer is
> >> newer than the revert of that change.
> >>
> >>  libstdc++-v3/include/std/format               | 19 ++++++++++++++-----
> >>  .../std/format/ranges/format_kind_neg.cc      | 15 ++++++++++-----
> >>  2 files changed, 24 insertions(+), 10 deletions(-)
> >>
> >> diff --git a/libstdc++-v3/include/std/format
> b/libstdc++-v3/include/std/format
> >> index bfda5895e0c0..887e891f2096 100644
> >> --- a/libstdc++-v3/include/std/format
> >> +++ b/libstdc++-v3/include/std/format
> >> @@ -5464,13 +5464,22 @@ namespace __format
> >>      debug_string
> >>    };
> >>
> >> -  /// @cond undocumented
> >> +  /** @brief A constant determining how a range should be formatted.
> >> +   *
> >> +   * The primary template of `std::format_kind` cannot be instantiated.
> >> +   * There is a partial specialization for input ranges and you can
> >> +   * specialize the variable template for your own cv-unqualified types
> >> +   * that satisfy the `ranges::input_range` concept.
> >> +   *
> >> +   * @since C++23
> >> +   */
> >>    template<typename _Rg>
> >> -    constexpr auto format_kind =
> >> -    __primary_template_not_defined(
> >> -      format_kind<_Rg> // you can specialize this for non-const input
> ranges
> >> -    );
> >> +    constexpr auto format_kind = []{
> >> +      static_assert(false); // Cannot use primary template of
> 'std::format_kind'
> >
> >
> > Shouldn't the comment actually be used as static_assert message
> >
> > static_assert(false, "Cannot use primary template of
> 'std::format_kind'");
> >
> > ?
>
> Ha, yes, obviously that would be better! :-)
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://gcc.gnu.org/pipermail/libstdc++/attachments/20250515/741607e5/attachment.htm>