Re: [dnsext] DS digest downgrade

[Matt McCutchen <[e-mail]>]

[Note, I made the initial allegation of this issue in the dane/keyassure
group.]

On Mon, 2011-23-21 at 23:13 -0000, George Barwood wrote:
> Ok, I see it now, section 3
> 
>    Validator implementations SHOULD ignore DS RRs containing SHA-1
>    digests if DS RRs with SHA-256 digests are present in the DS RRset.

I wasn't aware of this.  I did not think to check the specifications of
individual algorithms; I would have expected a specification that
modifies the validation requirements like this to update RFC 4035.  I
have submitted an erratum to this effect.

Ad-hoc SHOULD-level statements for validators are no way to achieve a
fundamental security property, and leaving the implications for the DNS
administrator as an exercise to the reader is no way to achieve
interoperability.  In my view, the issue cannot be considered resolved
until there is a single document that states the uniformity requirement
for DS records and a MUST-level requirement for validators in an
algorithm-general way.

-- 
Matt