Re: [stir] Review of: draft-ietf-stir-rfc4474bis-10

[Paul Kyzivat <[e-mail]>]

On 8/18/16 1:05 PM, Dave Crocker wrote:
> On 8/18/2016 9:48 AM, Peterson, Jon wrote:
>> Oh, we've allowed for multiple Identity headers for some time - the
>> problem is we don't really give any reason why we allow them or any
>> sense of what it might mean to have more than one. We need to say at
>> least something about that, though largely this is a matter of forward
>> compatibility with some future use cases that we know are under
>> consideration that we don't want to rule out.
>
>
> 1. In the abstract, there could be different identifiers that are
> authenticated.  That's probably not supposed to be allowed for a SIP
> environment, but since there apparently can be different identifiers
> used as input to validation, perhaps it is allowed.
>
> 2. There can be different authorities used to create the authentication.
> This permits trust assessment of a combination of entities making
> validity assertions.
>
> 3. There can be different technical details to the signature.  Different
> canonicalizations, different crypto algorithms, different keys.  All of
> these might affect survival through transit and/or ability of the
> recipient to process.

I think all of these are plausible and valid possibilities.

What this means is that any *prioritization* of these needs to be a 
decision made by the validator (perhaps differently for different 
validators in the same call), and not by the signers.

	Thanks,
	Paul