Re: [DNSOP] SVCB and HTTPS SvcParam multiple value order on the wire

> On 31 Jul 2020, at 19:46, Pieter Lexis <pieter.lexis@powerdns.com> wrote:
> 
> Hi folks,
> 
> I'm working on implementing SVCB and HTTPS in PowerDNS and I have some
> questions about the wire-format for the multi-value parameters like
> ipv{4,6}hint and alpn.
> 
> When there are multiple IP addresses in a hint, in what order should
> they be on the wire?

As they are entered.

> I would expect them to be ordered like an A/AAAA
> RRSet's RDATA to be sorted as specified in 4034 section 6.3 ("… are
> sorted by treating the RDATA portion of the canonical form of each RR as a
> left-justified unsigned octet sequence in which the absence of an octet
> sorts before a zero octet."). The draft says the hints are "an unordered
> collection", but it would be great to mandate an on-the-wire ordering
> here.

RFC 4034 is about validating a RRset. One needs to have a defined order
to DIGEST the RRset.  The order of the records on the wire is UNDEFINED.

> This will only work, of course, if multi-valued SvcParams are a set
> (where duplicates are disallowed/ignored), which is also not explicit in
> the draft for ipv{4,6}hint and alpn.
> 
> For the "mandatory" key, a sensible ordering (ascending) is specified
> and it is explicit that a key can only be present once in the set.
> 
> Cheers,
> 
> Pieter
> 
> -- 
> Pieter Lexis
> PowerDNS.COM BV -- https://www.powerdns.com
> 
> _______________________________________________
> DNSOP mailing list
> DNSOP@ietf.org
> https://www.ietf.org/mailman/listinfo/dnsop

-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742              INTERNET: marka@isc.org