On Tue, Jan 12, 2016 at 9:44 PM, Viktor Dukhovni <[email protected]>
wrote:

> >
> > On Jan 12, 2016, at 9:32 PM, John R Levine <[email protected]> wrote:
> >
> > They've appeared under transport names for the past 15 years, which
> means that people have expectations about how they're used which I would
> not casually ignore.
>
> And yet client identities are not really transport-specific.
>
> >
> > There are a lot of other prefixed names floating around the DNS.  If
> someone attempted to use a client name like _spf or _sip or _domainkey or
> _dmarc or _adsp or _vouch they and their users would experience an eternity
> of pain from name collisions.
>
> Only if there are TLSA records for those names.
>

Right.

Also recall that the proposed owner name is: _service.[client-domain-name].
So a zone operator can define client domain name structures in a way that
can address any namespace collision issues they wish to avoid. Presumably,
an "_spf.device1.dept.example.com" TXT record would be about SPF rules
pertaining to device1.dept.example.com, so there is likely not an issue with
it co-existing with a client TLSA record at that same name.

-- 
Shumon Huque
_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane

Reply via email to