[ietf-dkim] over-the-wire (in)compatibility between pre-IETF DKIM and (eventual) IETF DKIM

[Stephen Farrell]

Michael Thomas wrote:
> Stephen Farrell wrote:
> 
>> I thought changing the c14n actually was agreed? Changes to
>> the signature construct would appear to have some support on the
>> list. Formally of course none of these changes are agreed
>> since we're not yet a wg.
> 
> Sorry -- my decoder ring didn't decode c14n and I thought
> it was some actual canonicalization algorithm. Yes, I think
> we've agreed to that.

That's the main point here - if we change the c14n alg then
unmodified verifiers cannot validate those signatures and
therefore some other changes can be made (if agreed, useful,
etc., etc., etc.) without any additional impact on compatibility.

You can validly argue against any such change which is proposed,
but not, I beieve on compatibility grounds - if the proposal
is such that it makes no difference given the already agreed
c14n change.

The case in point (as an example not a recommendation) is
that disussed in the signature construct thread (which is
not "agreed" but does at least have some proponents).

I really do understand the reluctance to change in general
but at least in terms of signature construct compatability I
think that a consensus to change the c14n alg means an
acceptance of signature construct incompatability for our
"best" signatures.

Hopefully that horse is now sufficiently well flogged.

Stephen.