[dns-operations] RES_USE_INET6 and the inet6 resolver option

[Paul Vixie]

At the time, the advanced sockets API Siegfried this containerization. I was just supporting that spec. It's all dead now and should be removed.

On Feb 8, 2019, 02:23, at 02:23, Florian Weimer <fweimer at redhat.com> wrote:
>I want to remove this feature from glibc.
>
>This configuration option predates getaddrinfo (the API from RFC 3493)
>and gethostbyname2.  With RES_USE_INET6, the behavior of the legacy
>gethostbyname IPv4 address name lookup function changes so that instead
>of a 4-byte address, it returns a 16-byte address.  struct hostent has
>a
>length field, so this is in theory supported by the interface. 
>However,
>any caller will still have to be changed manually, to fill a struct
>sockaddr_in6 from it instead of a struct sockaddr_in.
>
>The result is that many applications simply break if this option is
>active (instead of being able to use IPv6 without changes, as may have
>been the intent), so it does not seem particularly useful to have this
>option available.  Since it also results in some awkward layering
>violations (why would host name lookups in an LDAP database care about
>resolv.conf settings?), I've deprecated it a while back and think it's
>now time to remove it.
>
>Comments?  Is anyone using this?
>
>Thanks,
>Florian
>_______________________________________________
>dns-operations mailing list
>dns-operations at lists.dns-oarc.net
>https://lists.dns-oarc.net/mailman/listinfo/dns-operations
>dns-operations mailing list
>https://lists.dns-oarc.net/mailman/listinfo/dns-operations
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dns-oarc.net/pipermail/dns-operations/attachments/20190208/d990e001/attachment.html>