On Wed, 1 Apr 2015, John R Levine wrote:
I do not understand the advantage of base32 in the QNAME.
As Viktor pointed out, the advantage is that the server can easily recover
the local-part from the query, which makes it possible for a specialized
server to do whatever it does and generate a response dynamically. You can't
do that with hashes.
Which might make sense for SMTP servers or whatever new protocol servers
you come up with, but not for DNS QNAMEs.
I've written DNS servers that generate responses on the fly from a database
where it does application-specific lookups and transformations. It's
surprisingly easy. They don't do DNSSEC yet but I'm planning to take a whack
at that later this year.
And you will be re-creating user/zone enumeration walking if you would.
And you would violate the rule that only mail servers apparently may
interpret mailbox names, and not secondary name servers. Unless you add
a whole bunch of other requirements to this document, like DNS servers
need to synchronise mailbox names and DNS servers must require online
DNSSEC signing.
Anything more complicated than a simple query-response, does not belong
in the DNS. Once you are not in the DNS, your application can use the
original LHS without any encoding and use whateve non-DNS protocol
extension you come up with.
Paul
_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane
