[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


To: Derek Atkins <warlord@MIT.EDU>
Cc: Edward Lewis <lewis@tislabs.com>, Simon Josefsson <jas@extundo.com>, <dnssec@cafax.se>
From: Jakob Schlyter <jakob@crt.se>
Date: Fri, 31 Aug 2001 17:29:29 +0200 (MEST)
Delivery-Date: Fri Aug 31 20:35:34 2001
In-Reply-To: <sjmu1yotjqf.fsf@rcn.ihtfp.org>
Sender: owner-dnssec@cafax.se
Subject: Re: CERTificates and public keys

On 31 Aug 2001, Derek Atkins wrote:

> There is absolutely no assumption at all in RFC 2538 that a
> certificate is signed.  If there is, please point out the section so I
> can go re-read it again.

rfc 2538 section 1 - "A certificate is a binding, through a cryptographic
digital signature, of a public key, a validity interval and/or conditions,
and identity, authorization, or other information."


> This means that when you "enter a key into a certificate" you don't
> have to worry about who signs it, because quite frankly NOBODY has to
> "sign" it.

I think we should separate certificates, that contains its own signature
(which is probably not a all related to dnssec signatures) and simple
public keys (that needs and depends on dnssec to be trusted).


> The main reason, IMHO, to use a CERT record over a KEY record is that
> KEY records imply DNSSec keys whereas CERT records imply application
> keys.

one of the the problems are that KEY is used both for storing keys used by
the infrastructure itself and for applications. I believe the solution is
to create a separate application key rr-type as described in the appkey
draft (draft-schlyter-appkey-00.txt).


	jakob


Home | Date list | Subject list