[ietf-dkim] is s= really needed?

[Stephen Farrell]

Michael Thomas wrote:
> Paul Hoffman wrote:
> 
>> At 11:30 PM +0100 7/1/06, Stephen Farrell wrote:
>>
>>> 3.6.1, "s=". Given that new tags can be added to key records
>>> without affecting compliant implementations, I don't see why this
>>> is needed now and suggest deleting it.
>>
>>
>> Agree, but could also be happy leaving it in for some odd extension.
> 
> I'd say that it's better to leave it in than not because if new records are
> added for different purposes (eg s=espresso-over-ip) a compliant DKIM
> will not allow those keys use for smtp, whereas if we extended it later
> they would.

Ah, ok - so the intent is that I don't ignore "s=foo" if I know that
"s=email" exists. I'd read it as saying that verifiers could ignore
that too but I guess your interpretation makes more sense. Maybe adding
a sentence like "Verifiers that expect a particular value for this
tag MUST NOT use records containing some other value."

S.