Re: [mif] [DNSOP] [dnsext] 2nd Last Call for MIF DNS server selection document

[Keith Moore <[e-mail]>]

On Oct 21, 2011, at 11:07 AM, Ted Lemon wrote:

> On Oct 21, 2011, at 3:15 AM, <[e-mail]>
>  <[e-mail]> wrote:
>> There could perhaps be another draft, which would say that if name is "foo"
>> it should not be appended with search lists but "foo." might? And whatever
>> other differences in their handling would be, and what impacts it would have
>> e.g. intranet designers?
> 
> I tend to agree with others who have observed that this question is beyond the WG's core competency.  But there really is a mif question having to do with how search lists are handled.   Personally I tend to side with the crowd that believes that DNS search lists should be deprecated with extreme prejudice, but if the consensus is otherwise, I think this draft you describe does need to be written.
> 

IMO: search lists are useful, but only with "bare names" - and the behavior of those should be implementation dependent.  Trying to nail it down will break too much widespread practice.

Names containing "." should not be subject to search lists.  Given a name like foo.bar, there's no reliable way to tell whether "bar" is a TLD or a subdomain of something in the search list. 

(No, trying to look up "foo.bar" starting from the DNS root, and failing over if that lookup fails, is not sufficient, because of (a) temporary failures and (b) it still doesn't tell you what the user _meant_.  It makes much more sense to say that any name containing a "." is unambiguously an FQDN.)