Re: [Dcrup] FW: IETF WG state changed for draft-ietf-dcrup-dkim-usage

[Hector Santos <[e-mail]>]

On 8/14/2017 9:33 PM, John Levine wrote:
> In article <8695284.qrNCWkNy01@kitterma-e6430> you write:
>>> Another way to look at this: I think it's more appropriate to render
>>> rsa-sha1 obsolete, but this approach seems as if we want to act like it
>>> never existed.
>>
>> Fast forward a few years:  Is the fact that it ever existed relevant to
>> anything?  I think it's highly unlikely.
>
> The Internet being the Internet, sha-1 hashes will trickle in forever.
> I'd rather have the diagnostic say "obsolete hash" than "syntax
> error."
>
> If you look at this curdle draft that deprecates RC4, it goes through
> and makes changes to turn OPTIONAL to MUST NOT and the like, but it
> doesn't try to undefine the obsolete rc4 crypto modes.  I think you'll
> find that typical in crypto updates.

+1.  The odds are very good that it will continue to be a "will not 
sign, but will verify as needed" behavior for a very long time.  And 
probably, local rules will apply on a site by site basis using other 
bits in the mail.   I am not about to break communications with my 
existing customers or break communications for my installation base.

My key concern here is that the IETF guidelines should not change 
history for future developers. It should help developers as it will be 
a big surprise when SHA1 mail is encountered and these new developers 
are not ready for it.  The 20/20 hindsight question will be raised:

     "Why wasn't it documented that DKIM mail with SHA1 will exist and
      developers should be prepared to deal with it?"

Having the IETF suggest that sha1 DKIM mail MUST|SHOULD be rejected 
with a 55z policy code, is probably not good advice -- IMO.

-- 
HLS