[ietf-dkim] Proposal: get rid of x=

[Paul Hoffman]

At 9:12 PM -0700 4/11/06, Dave Crocker wrote:
>>>>Further, section 6.4 makes no sense and has to be eliminated or 
>>>>seriously re-written. You can't put a header in a message for a 
>>>>fact that will become untrue in the future.
>>>
>>>The header simply says that the
>>>message was validated.  Not that it can be validated at some point 
>>>in the future.
>>
>>There is a huge disconnect here. x= is *not* talking about the 
>>ability to validate at some point in the future; it talks about a 
>>message that is valid at one point becoming invalid at a later 
>>point.
>
>It should talk about being able to conduct a validation within a 
>window of time, and not being able to do it after the window closes.

If that's what the WG wants, great. But that is *not* what the document says:
        Signature expiration in seconds-since-1970 format
        as an absolute date, not as a time delta from the signing
        timestamp.  Signatures MUST NOT be considered valid if the
        current time at the verifier is past the expiration date.

>This is not about a "contract signature" becoming invalid.

"MUST NOT be considered valid" sure sounds like "becoming invalid" to me.

>   It is more like a traffic light changing.  Transit is ephemeral, 
>so it should not be surprising that a mechanism related to transit 
>is ephemeral.

This is a good analogy, one that it seems many people in the WG want. 
But it is not what is in the document.

>>         sending domain publishes an authentication policy of some kind,
>>         and the message passed the authentication tests
>>Note the past tense used: "passed the authentication tests". In a 
>>normal environment, that is sufficient for a MUA to give a sensible 
>>notice. But in an environment where a message can be valid at one 
>>moment and invalid at the next, that is not sufficient to tell the 
>>MUA what to display at any particular time.
>>
>>Is this clearer?
>
>"passed the authentication tests" is an accurate description of what 
>took place.  "Message valid at one moment and not at the next" is 
>not.

We disagree, then. If the verifier checks the validity and it passes, 
and later checks the validity and it fails because of x=, then 
"Message valid at one moment and not at the next" seems to be a 
reasonable technical description of what happened.

>A DKIM signature says that someone asserts that they are accountable 
>for message transit.

And here we are fully agreeing again, although no such simple 
statement exists in the document. (Hint: search for the word 
"responsible".)

>You are confusing limitations in the ability to perform a validation 
>check, with the continuation of the assertion's validity.

No, I'm not. x= says *nothing* about "limitations in the ability to 
perform a validation check"; it *does* talk about "continuation of 
the assertion's validity".

>If you go through an intersection when the light is green (for your 
>direction) it was valid for you to proceed.  The light changes.  The 
>validity of your having transited the intersection does not.

And, again, we fully agree.