Hey Paul,

(with the usual apologies for the MIME-crime that follows)

> On Jan 24, 2018, at 20:50, Paul Vixie <[email protected]> wrote:
> 
> Mark Andrews wrote:
>>> On 25 Jan 2018, at 8:38 am, Paul Vixie<[email protected]>  wrote:
>>> 
>>> viktor, i don't disagree with your goals, but i have a proposal as to 
>>> method.
>>> 
>>> no resolver should be sending single-label names in DNS requests, period.
>> 
>> I would qualify that as gethostbyaddr / getaddrinfo etc. should not be ….
>> 
>> Resolvers do more than lookup host names.

I was about to point out that in a QNAME-minimised future, every legitimate 
query received by a TLD nameserver has a single-label QNAME. But now I don't 
have to.

> as others pointed out off-thread, you are right, and that's what i meant.

If I agree that this is a worthy goal (and I think I do, although I haven't 
thought it through in context very well) the obvious question is whether it's 
possible to get there and, if so, how.

There is an awful lot of entrenched practice (including every corporate 
environment I've ever worked) where a search list and single-label hostnames 
are either a convenient short-cut or an absolute requirement, like it or not. 
The collateral damage that would result from slamming that door in enough 
operating environments to make a difference is surely greater than customers, 
managers and shareholders would tolerate.

What am I missing?


Joe

_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to