Re: [Dcrup] Deprecating algorithms

["Murray S. Kucherawy" <[e-mail]>]

On Tue, Jun 13, 2017 at 6:37 PM, Murray S. Kucherawy <[e-mail]>
wrote:

> On Tue, Jun 13, 2017 at 7:52 AM, Martin Thomson <[e-mail]>
> wrote:
>
> This is a fine thing.  Do you want to propose text for the charter?
>>
>
> The diff to the current charter is going to be mighty small, so sure, I'll
> do it this week sometime.
>

Proposed revised charter text.  A crypto person should review it to make
sure it's not nonsense since I only know enough about crypto to be
dangerous.  Apologies for the silly wrapping; I fought with the gmail
interface to make it pretty and failed.

-- snip --

The DKIM Crypto Update (DCRUP) Working Group is chartered to update
DomainKeys Identified Mail (DKIM, RFC 6376) to handle more modern
cryptographic algorithms and key sizes. DKIM (RFC 6376) signatures
include a tag that identifies the hash algorithm and signing algorithm
used in the signature. The only current algorithm is RSA, with advice
that signing keys should be between 1024 and 2048 bits. While 1024 bit
signatures are common, longer signatures are not because bugs in DNS
provisioning software prevent publishing longer keys as DNS TXT records.

DKIM also currently supports use of SHA1 coupled with RSA.  SHA1 has been
formally deprecated due to weakness especially when used in the context
transport security, though the risk of a successful preimage attack is less
severe.  Still, the community wishes to discourage its continued use in the
DKIM context.

DCRUP will consider four types of changes to DKIM: additional signing
algorithms such as those based on elliptic curves, changes to key
strength advice and requirements, deprecating the use of SHA1, and new
public key forms, such as
putting the public key in the signature and a hash of the key in the
DNS to bypass bugs in DNS provisioning software that prevent publishing
longer keys as DNS TXT records. It will limit itself to existing
implemented algorithms and key forms. Other changes to DKIM, such as new
message canonicalization schemes, are out of scope. The WG will as far
as possible avoid changes incompatible with deployed DKIM signers and
verifiers.