> On Nov 23, 2017, at 11:13 AM, dnsop-requ...@ietf.org wrote: > > Even so, I know that at least one CA has received enough complaints from > customers with REFUSED private domains that they have already updated > their implementation to permit certificates in unresolvable zones that > lack DNSSEC. It worked before CAA and I don't think there's any particular > advantage to breaking it.
Supporting private domains that (really) fail to resolve makes things more complicated and perhaps needlessly so. Perhaps the onus should be on the operators of the zones in question to make appropriate changes. In the mean time, if REFUSED and the like also need to be supported, then the SOA lookup strategy can be modified to walk up the tree in the face of (real) lookup errors, to conclude that a subdomain is insecure once an ancestor domain is found insecure. It is unfortunate that we routinely jump through hoops to work-around poor practices by others, and keep finding more ways to do so as time goes by. Is it a good idea to enshrine such a private domains hack as standard practice? Or can we grow some spine and require domains that want certificates (say starting January 2019) to not fail public DNS lookups? That is, perhaps the present work-arounds should have a sunset date. -- Viktor. _______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop