Just a small clarification,

*"I'm going to ignore the fact that this is a core ABI issue and therefore
cannot ever be standardized ..."*

If an implementation consistently uses the same calling convention for
member functions as it does free functions then the cast can just return
the address of the hidden free function that already exists
If an implementation inconsistently uses a different calling convention for
member functions as it does for free functions then the cast can just
return the address of a perfectly forwarding trampoline function.
In either case, no ABI changes are needed.



On Sun, Jan 14, 2024 at 6:55 PM Jason McKesson via Std-Proposals <
std-proposals_at_[hidden]> wrote:

> On Sun, Jan 14, 2024 at 6:28 PM Frederick Virchanza Gotham via
> Std-Proposals <std-proposals_at_[hidden]> wrote:
> >
> > On Sat, Jan 13, 2024 at 7:04 PM Jarrad Waterloo wrote:
> > >
> > > I really think this needs to be a language feature that is portable.
> >
> >
> > I agree that it should be a portable language feature in C++26,
>
> What is "it"? Because the two of you are talking about two
> *fundamentally* different things here.
>
> You want to get a free function pointer that behaves exactly like the
> memer pointer does. Jarrad's proposal *is not that*.
>
> Jarrad's proposal is to convert a member function pointer into a free
> function pointer that calls *exactly* the member function that was
> specified in the expression that extracted the member pointer. That
> is, no virtual dispatch.
>
> These aren't the same thing. You would not get what you want from
> Jarrad's proposal and vice-versa.
>
> > And while I'm talking . . . maybe it would be worth mentioning that
> > name mangling should be standardised too?
>
> I'm going to ignore the fact that this is a core ABI issue and
> therefore cannot ever be standardized and instead ask a different
> question. From a standard perspective... what is "name mangling"?
> Like, what would it mean?
>
> Name mangling is an implementation detail used to allow ABIs designed
> for C to work with C++ function overloading, namespaces, and the like.
> That's it. C++ has features that allow two entities to have the same
> name yet are different. C ABIs can't handle that. Name mangling is how
> they square that circle. It is completely invisible to a user writing
> standard C++.
> --
> Std-Proposals mailing list
> Std-Proposals_at_[hidden]
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>