Re: draft-arends-dnsnr-00

[Roy Badami <[e-mail]>]

>>>>> "Roy" == Roy Arends <[e-mail]> writes:

    Roy> Ben. It is very simple:

    Roy> If _you_ state that _you_ signed record type AAAA and record
    Roy> type MX for a given name, while not actually signing record
    Roy> type AAAA and record type MX, that would be violating the
    Roy> spec.

I'm going to have to agree with Ben here.  For it to consitute
non-repudiation as commonly understood, then doing as you describe
would have to be provable by a third party by means of the
cryptographic protocol, not merely a violation of the specification.

Non-repudiation of signing would mean that if you signed something you
can't later deny that you've signed it.  Or rather, that if you try, a
third party can prove that you're lying.  But that's a defining
property of signing and noone would explicitly state it; anyone can
check the signature and prove that you did indeed sign the data.
(Here "you" actually means someone who has access to your private key,
but that's an implicit assumption in these things.)

But that's not what we're taling about, I think.  What we're talking
about is signing some statement (some kind of NSEC++ record) that
asserts that a record doesn't exist.  It's just signing a negative
statement; it's not asserting that you didn't sign something; and it
has nothing to do with either non-repudiation or repudiability.

Indeed (not that it's relevant to the argument) but you *might* have
signed such a record in the past, that doesn't mean you can't now sign
a statement to the effect that the record no longer exists.

I am not a cyptologist, but AIUI, non-repudiation means that some
other party (either the other party in the transaction, or some third
party) is able to prove that the transaction did take place.

FWIW, the converse, repudiability, means that if you and I engage in a
cryptographic transaction, no third party observer (and perhaps even
neither of us) has evidence that would prove that the transaction
actually took place.

I really don't see either non-repudiation or repudiablity as being
relevent to the protocols we're discussing here.  We're simply talking
about making a signed statement that a particular record does not
exist.

    -roy

--
to unsubscribe send a message to [e-mail] with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/namedroppers/>