Re: [sidr] I-D Action: draft-ietf-sidr-algorithm-agility-01.txt

[Sandra Murphy <[e-mail]>]

On Mon, 11 Jul 2011, Stephen Kent wrote:

> At 1:44 PM -0400 7/11/11, Sandra Murphy wrote:
>>>> ...
>>> 
>>> Agreed. The CP cites the alg spec (draft-ietf-sidr-rpki-algs). However, 
>>> this doc say that the alg specs doc will be updated to reflect the new alg 
>>> suite, and to include the timeline for the alg transition. Once that 
>>> happens, a failure to comply with the alg transition procedure described 
>>> here will imply noncompliance with the CP.
>> 
>> S---T---R---E---T---C---H???
>> 
>> If the non-compliance with this draft was to fail to update the algs 
>> document, then the failure to comply with the procedure would not imply 
>> non-compliance with the CP.
>> 
>> --Sandy, speaking as wg chair
>
> Sandy,
>
> But stretching is the usual pre-exercise warm up, and the transition to a
> new alg suite will be an exercise, so ... :-).
>

Yes, especially when the exercise involves significant muscle movement, 
and alg transition involves significant movement of the brain muscle, 
so...  :-)

> Stated less circuitously, the CP currently mandates support for the algs
> in the Alg Spec. These algs used to be in the CP, but it was decided to
> move them into a separate doc, to avoid the need ti change the CP when the
> algs change. An early version of the alg transition doc called for updating
> the CP to reflect alg transition. But, we moved the algs spec to a separate
> doc, so the alg transition now cals for the alg spec to be replaced with
> a new doc that calls out the new algs and provides the transition timeline.
>
> We have gone down the path of document modularization and indirection, this 
> is where we wound up!

I understand the reasoning (I can stretch that far).

But the desired outcome is for any violation of the transition procedure 
to violate the CP.  Right?

As you said, this follows from the first step of the procedure, which is 
to update the alg document with the transition timeline, thereby 
indirectly updating the CP with the transition.  Subsequent violation of 
the transition procedure would therefore violate the CP.

But if the violation of the transition procedure is to fall at the first 
hurdle - to fail to update the alg document with the transition timeline, 
then any subsequent violation of the procedure would no longer violate the 
CP, even indirectly.

Right?

--Sandy, speaking as wg chair.


>
> Steve
>