Hi Daniel. A bit late maybe, but...
> At its core, there's a concern here that asking a server to bind to a > name might actually bind it to multiple addresses (since a name can have > multiple addresses). I'm not convinced that Net::Server can be properly > made to bind to multiple addresses with a single "host" configuration, > but i agree that for the sake of congruity with the rest of the perl > world, Net::Server should preferentially bind to the IPv4 address if a > given name maps to both an IPv4 and an IPv6 address. I don't agree with you (and "the rest of the perl world") here on a general basis. getaddrinfo returns any IPv6 addresses before IPv4 *by design*. If it were the other way around the (inevitable) transition to IPv6 would be much harder. Of course this is only true if asking for AF_UNSPEC (or the trivial AF_INET6) records. In that light rewriting Net::Server to always prioritize IPv4 is dead wrong and should be avoided. If selecting an IPv6 address is causing any problems, this is most likely caused by old IPv4-only mindset (like localhost always and only resolves to 127.0.0.1). If "we" are going to have success in the v4->v6 transition this mindset must be changed instead of breaking IPv6. I do support your rewrite to 5.14 core modules for "native" IPv6 functionality, but not this "do not break any (already broken) IPv4 applications". -- Pelle "D’ä e å, vett ja”, skrek ja, för ja ble rasen, ”å i åa ä e ö, hörer han lite, d’ä e å, å i åa ä e ö" - Gustav Fröding, 1895 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
