[Bug libstdc++/113099] locale without RTTI uses dynamic_cast before gcc 13.2 or has ODR violation since gcc 13.2

["andysem at mail dot ru" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113099

--- Comment #11 from andysem at mail dot ru ---
> I'm not sure what you mean by "the compiler is free to generate code that takes it into account." Takes what into account? What problem does that freedom cause?

I mean the compiler could move (some part of) dynamic_cast to precede the check
for the standard facet. I.e. of something like this:

  template< typename _Facet >
  const _Facet* __try_use_facet(locale const& loc)
  {
    const size_t __i = _Facet::id._M_id();
    const locale::facet** __facets = __loc._M_impl->_M_facets;
    const _Facet* __dyn_facet = __dynamic_cast< const _Facet* >(__facets[__i]);

    // checks for every standard facet type
    if (__is_same(_Facet, ...))
      return static_cast<const _Facet*>(__facets[__i]);

    return __dyn_facet;
  }

>> I think, a failing dynamic_cast would not be useful as this would make
>> std::use_facet unusable with -fno-rtti.
>
> I don't see a problem with that. If you want to use a polymorphic container of facets, you need RTTI. The standard facets will work, but it doesn't seem reasonable to expect it to work for program-defined facets.

AFAIK, the standard or libstdc++ docs do not require RTTI for std::locale to
function. In fact, the facet::id stuff seems to exist precisely to make RTTI
optional in the implementations. Besides, the code, as it was written, seems to
intend to work without RTTI by switching to static_cast instead of
dynamic_cast. So making std::use_facet always fail would go against that
intention.

> And a std::use_facet that fails seems better than one that segfaults, doesn't it?

No, not really. It still means users cannot use locale without RTTI, just for a
different reason.

> And I guess if the user's derived facet uses virtual inheritance from std::locale::facet, then this could break with the static_cast<std::ctype<char>*> in libstdc++ today. Hmm.

std::ctype non-virtually derives from std::locale::facet, so no, that would not
break. What wouldn't work is this:

   class my_facet : virtual public std::ctype< char > {};

   my_facet const& fac = std::use_facet< my_facet >(loc);

But this would fail in compile time with no RTTI, which *is* better than a
segfault or an exception in runtime.