Re: [DNSOP] Verifying errata 5316 against RFC1034.

[Mukund Sivaraman <[e-mail]>]

On Sun, Apr 01, 2018 at 01:33:17PM -0400, Warren Kumari wrote:
> Can folk help me understand what should happen with this errata?
> W

To elaborate further:

IMO there's no argument against caching if the cached record set (with
wildcard owner name) was not used in synthesis of RRs. I suspect RFC
1034 language was written to prevent that, not the actual caching
itself.

Caching takes place not just by BIND, but Unbound as well and does not
cause problems, so the stronger requirement is unnecessary and ought to
be re-worded.

That is.. unless there is something I've missed. ;) What else is the
problem in caching wildcard records?

> Note that if behaviors have changes, and implementations should now
> cache the record, then we need to document that in a -bis (or similar)
> document.

I agree about this in the scope that we even allow some synthesis with
DNSSEC now in RFC 8198.

But the errata is still for the 1987 document, i.e., as it was back in
time. There doesn't seem to be any sense in having this requirement and
implementations ignore it anyway.

		Mukund