https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124890
--- Comment #3 from Lénárd Szolnoki <leni536 at gmail dot com> --- > Compared to the forwarding reference approach this avoids needing to cast > __lhs to std::indirect throughout, while continuing to accept types derived > from std::indirect, but at the expense of no longer accepting types > convertible to std::indirect only via a conversion operator (and for which > std::indirect is an associated class). As this now observably deviates from the current draft, I think filing an LWG issue would be warranted.
