On Thu, May 16, 2024 at 3:24 PM Jason McKesson wrote:
>
> So you want the language to change in a way that can only be useful
> for compiler-specific code, in order to do something you *can already
> do* with the current language in a way that works on all compilers.
>
> This is not what "motivation" looks like.


Actually, rather than try to make a new 'bit_cast', the real remedy
here is for the Standard to mandate that an std::tuple shall be
trivially copyable if all of its elements are trivially copyable, and
so then we'll be able to use std::bit_cast on a std::tuple to pass it
into a consteval hash function.

I've done some web searching to try ascertain why std::tuple isn't
guaranteed by the Standard to be trivially copyable if all of its
elements are trivially copyable, and I've seen explanations along the
lines of "It's just too complicated to implement on top of the
existing implementations and it gets a bit hairy". Well I think I have
it working for libstdc++ by changing 21 lines in the <tuple> header
file, you can see my changes here:

    https://github.com/healytpk/tuple_trivially_copyable/commit/b274f779075d2c53664da0a159dbafa10d668f10

Tomorrow I'll try make a similar patch for the Microsoft
implementation of std::tuple. Then if I have a viable patch file for
both compilers, I think it would make sense to write a paper to
propose that std::tuple shall be trivially copyable whenever all of
its elements are trivially copyable. Same goes for std::pair.