On 9 Oct 2020, at 11:58, Roy Arends wrote:
> On 9 Oct 2020, at 10:38, Andrew McConachie <[email protected]> wrote:
Hi Roy and Joe,
It’s not clear to me whether the document is advising to only use
this facility when a sub-domain of a public domain name is
unavailable, or to optionally use this facility based on the user’s
preference. What I would like the document to say is that only when a
sub-domain of a public domain is unavailable should this facility be
considered. The reader should get the impression that they should try
really really hard to not use the ISO-3166 reserved string if they
can.
This is marked as a BCP and so I would expect to see this advice
prominent in the document. Since, IMO at least, that is the best
current practice. Only when a user cannot use a sub-domain of a
domain they control should they even consider using the ISO-3166
reserved string. Ideally there could be a new section discussing this
advice between the current sections 1 and 2. That way the reader will
encounter the best practice before encountering the work around.
Thanks for your comment Andrew,
The next version will contain more text directed to this.
IMHO, the mere availability of a subdomain (of an existing domain)
should not automatically preclude the use of a private top-level
domain. That is, I disagree with the notion that “they should try
really really hard to not use the ISO-3166 reserved string”.
Note that a domain may not always be tied to the same legal entity.
When software or devices are shipped with a default configuration that
is meant to work only locally (there are a few scenarios that include
home use or corporate use), using a public domain is problematic.
Queries will leak to the authoritative servers of that public domain,
long after the public domain has changed hands.
It is also not desirable from a legal and even operational standpoint
if software that is shipped with a default public domain will be
“phoning home” all the time.
There are likely to be many different use cases, some where a public
domain is useful, some where a private-use domain is useful.
Hi Roy,
Thanks for the response. I think we’re basically in agreement here.
If the mere availability of a sub-domain (of an existing domain)
precluded ever using this ISO-3166 reserved string then there would
never be a valid use case for the ISO-3166 string. If only because one
can always register a new domain and delegate a sub-domain. We can
quibble over what “try really really hard” means, but it would not
improve the document to enumerate every possible use case and proscribe
which route the user should take for each one.
I look forward to your next version with more text on this topic.
Thanks,
Andrew
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop
