rrset-roundrobin

[George Thessalonikefs]

Hi all,

I want to thank you for the input.
It seems that the consensus is to change the default to yes and thus
enforcing the rrset definition of an unordered set.

@Shumon Huque:
Nice suggestion. The dnsop thread from a couple of years ago
[https://mailarchive.ietf.org/arch/msg/dnsop/_diK61t7YmZu_DJvRvUAhvKVSh4/]
was an interesting read.
On your concern about side channel attacks, it is a rotation scheme but
unbound rotates the rrset randomly by using client data (query ID) and
not state on the resolver side; also mentioned in that thread.

@Gavin McCullagh:
All changes per release are communicated to users(this list), package
maintainers (another list), a news item on our website and tweets
linking to said news article. It is then up to the users/maintainers how
to proceed with the changes.

Best regards,
-- George