Re: [dane] Digest Algorithm Agility discussion (checkpoint)

[Viktor Dukhovni <[e-mail]>]

On Wed, Mar 19, 2014 at 10:08:38AM -0400, Scott Rose wrote:

> On Mar 18, 2014, at 4:00 PM, Viktor Dukhovni <[e-mail]> wrote:
>
> > My sense is that regardless, there is not much enthusias for
> > negotiating a single digest based on what digests the server offers,
> > with the client choosing its most preferred one.
> > 
> > Is this an accurate summary of the group's consensus view?  Does
> > anyone want to defend the view of TLSA digests as a menu of options
> > from which the client can choose one?
>
> Don't know about the rest of the WG, but it's mine.  Some communities have
> a larger local policy that they want to enforce, and the client will prefer
> that primarily, with potential fallbacks.

Sorry, could you confirm the meaning of the above sentence?  Probably
my fault, but I am not 100% sure whether you're saying that clients:

    - should (proposed agility protocol)
    - may (employ adaptive local policy that amounts to the above), or
    - must not

ignore lesser ranked (by the client's local policy) digest matching
types in the server's TLSA RRset.

> > If not, I will drop the digest agility portion of the SMTP draft.
> > In the OPs draft we can encourage server operators (SHOULD) to
> > apply all digests equally to all objects, because that's more robust
> > in the face of local policy, and results when this is not done may
> > not be what the operator wanted.
> 
> Still need to see the final text, but looks like it is the right direction.

This seems to suggest a vote for either "may" or "must not" above...

-- 
	Viktor.