Hoi Paul, On 7 Nov 2026, at 06:57, Paul Wouters <[email protected]> wrote:
> This seems fine, but it should probably also clarify where RRSIG RRs go for > those CNAMEs. The case we were reacting to (and trying to prevent recurrence of) was a response supplied to stub resolvers that had critical expectations about the ordering of records in the answer section. The cases we know about come from a long tail of legacy devices that might never be upgraded, hence trying to do something about it in the spec. But I don't think we see such devices send queries with DO=1. We can check, of course. But it feels like many of these cases might even pre-date DNSSEC. If there are no known negative consequences of existing practices regarding RRSIG packing in the answer section, it feels like it would be over-specifying to require something new? > Or possibly, in general, especially if the goal is that a resolver can read > top down while , reading, validating and adding to its cache. This seems like a broader mandate than this draft has, if I recall earlier conversations about it correctly. Joe _______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
