Re: [keyassure] Opening issue #21: "Need to specify which crypto

[Sean Turner <[e-mail]>]

On 3/8/11 2:20 AM, Jakob Schlyter wrote:
> On 8 mar 2011, at 07.23, Brian Smith wrote:
>
>> It would be better to have both SHA-384 mandatory for clients. Those two algorithms are already practically required because they are both required for a good implementation of TLS 1.2. Making SHA-384 mandatory for clients makes it easier for server administrators to switch from SHA-256 to SHA-384 if that would ever be helpful; that could be a real possibility considering the deployment of SHA-3 on clients is a long way off.
>
> +1 - having two algorithms (although from the same family) from the start is better than just one. And we add SHA-3 and friends later - no need to "reserve" code points for them at this point.

I think I'm in this camp.  It's hard to use 2119-language for something 
that nobody can claim conformance to.  Somebody is still going to need 
to write a draft that says here's the SHA-3 alg, here's it's reference 
and we're updating the registry regardless so I think the draft could 
just as well do this:

    Value        Short description       Ref.
    -----------------------------------------------------
    0            No hash used            [This]
    1            SHA-1                   NIST FIPS 180-3
    2            SHA-256                 NIST FIPS 180-3
    3            SHA-384                 NIST FIPS 180-3
    4-254        Unassigned

    Applications to the registry can request specific values that have
    yet to be assigned. Note that there is every intent to update this
    registry when [SHA-3] is finalized.  Implementors are strongly
    encouraged to support a flexible implementation strategy in order
    to support an orderly migration to [SHA-3] shortly after this
    registry is updated.

or something like that.

I can already envision the GEN-Art comment/IESG discuss on needing a 
stable reference for SHA-3 if it's a normative reference.  If there's no 
requirements associated with it you'll probably get away with an 
informative reference to [SHA-3]:

[SHA-3] "Cryptographic Hash Algorithm Competition", NIST Computer
         Security Resource Center Cryptographic Hash Project.

There is precedent to using this reference but it was in a draft with an 
intended status of informational and it was an informational reference.

spt