On Mon, Apr 27, 2015 at 12:23:40AM +0100, Chris Monteiro wrote:

> Right, so reading up on TLSA, I can see how the port, certificate and
> certificate metadata are defined together and obviously I think this
> is a great implementation to kill of CAs as we know them.

This is the DANE working group mailing list, I think this is off
topic.

> However, I don't see why the TLSA syntax must require the preferred
> port AND the certificate hash as well. A systems administrator want to
> only specify use of say 443/TLS whilst relying on the existing CA
> system for certificate validation. Sure this is not where we want to
> be long term, but allowing to implement partial TLSA features would
> simplify things wouldn't it?

The administrator can use DANE-TA(2) or PKIX-TA(0) (whichever is
better applicable to the application protocol in question).  And
then need not publish the particular server certificate.

> Therefore I believe my proposal is to allow a cut-down implementation
> of TLSA for this purpose.

No, you're just new to the issues, and have not thought them through
yet.  Take your time.  You can follow-up on [email protected] if
you like, that's a user forum, not an IETF WG list.

-- 
        Viktor.

_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane

Reply via email to