On Thu, 12 Feb 2026 at 21:09, François Dumont <[email protected]> wrote:
> I had started the same on my side for unordered containers, I'm > attaching my WIP (untested). > > As you'll see my main concern was to rely on members protected > themselves with a feature test macro like > __glibcxx_generic_unordered_lookup. I guess that if you did so it means > that I'm too caution and that as long as > __glibcxx_associative_heterogeneous_erasure is defined then all the > others too. > Yes, because generic_unordered_lookup is defined for C++20 and later, and associative_heterogeneous_erasure is defined for C++23 or later, so if the latter is defined then the former must be.
