On Sat, 18 Mar 2023 at 00:49, Nathaniel Shead wrote:

> On Sat, Mar 18, 2023 at 7:36 AM Jonathan Wakely via Libstdc++
> <libstdc++@gcc.gnu.org> wrote:
> >
> > Tested x86_64-linux. Pushed to trunk. gcc-12 backport needed too.
> >
> > -- >8 --
> >
> > libstdc++-v3/ChangeLog:
> >
> >         PR libstdc++/109165
> >         * include/std/coroutine (hash<>::operator()): Add const.
> >         * testsuite/18_support/coroutines/hash.cc: New test.
> > ---
> >  libstdc++-v3/include/std/coroutine            |  2 +-
> >  .../testsuite/18_support/coroutines/hash.cc   | 22 +++++++++++++++++++
> >  2 files changed, 23 insertions(+), 1 deletion(-)
> >  create mode 100644 libstdc++-v3/testsuite/18_support/coroutines/hash.cc
> >
> > diff --git a/libstdc++-v3/include/std/coroutine
> b/libstdc++-v3/include/std/coroutine
> > index f6e65566f10..b0ca18949db 100644
> > --- a/libstdc++-v3/include/std/coroutine
> > +++ b/libstdc++-v3/include/std/coroutine
> > @@ -345,7 +345,7 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
> >      struct hash<coroutine_handle<_Promise>>
> >      {
> >        size_t
> > -      operator()(const coroutine_handle<_Promise>& __h) noexcept
> > +      operator()(const coroutine_handle<_Promise>& __h) const noexcept
> >        {
> >         return reinterpret_cast<size_t>(__h.address());
> >        }
> > diff --git a/libstdc++-v3/testsuite/18_support/coroutines/hash.cc
> b/libstdc++-v3/testsuite/18_support/coroutines/hash.cc
> > new file mode 100644
> > index 00000000000..68e5e640477
> > --- /dev/null
> > +++ b/libstdc++-v3/testsuite/18_support/coroutines/hash.cc
> > @@ -0,0 +1,22 @@
> > +// { dg-options "-std=gnu++2a" }
> > +// { dg-do run { target c++2a } }
> > +
> > +#include <coroutine>
> > +#include <testsuite_hooks.h>
> > +
> > +void
> > +test01()
> > +{
> > +  std::hash<std::noop_coroutine_handle> h;
> > +  std::size_t v = h(std::noop_coroutine());
> > +
> > +  const auto& ch = h;
> > +  std::size_t v2 = h(std::noop_coroutine()); // PR libstdc++/109165
>
> Is this supposed to be `std::size_t v2 = ch(...)`?
>
>
Doh! Yes, thanks. Fixed by the attached change.

Tested x86_64-linux, pushed to trunk.