Also for the record I've now gotten AFKMud to compile under Cygwin again (damn libraries) and confirmed that even on my Windows 10 Home PC that the code bound both addresses and accepted a connection from another PC in the house. Neither of which get assigned IPv6 addresses because our router is so backward it doesn't even support it at all.

I don't think portability is an issue here. It's just a matter of getting the DNS resolver process to play along now.

Fair enough.

I'll agree to disagree on portability, since I thought the principle argument against a dual stack socket was that not all (legacy) OS would support it, instead requiring an approach with multiple sockets (like the RFC was showing). That said, it's not my base to maintain and I suppose few people will try and compile and run this on hardware that ancient. Guess we'll find out if they do, should they find themselves here asking for support.