Re: [Rats] Where Do Interaction Models Go?

[Thomas Fossati <[e-mail]>]

Hi Henk,

On 23/07/2020, 14:08, "Henk Birkholz" <[e-mail]>
wrote:
> Hi list,
>
> our first virtual session on July 28th is approaching. At the last
> meeting, we highlighted the question: where do interaction models go?
>
> Context: there are three quite common interaction models (plus minimal
> requirements and information elements for them) that can be used to
> convey RATS Conceptual Messages, such as Evidence or Attestation
> Results. Most Conceptual Messages should include only up-to-date
> content (commonly referred to as "fresh") to be feasible. As a result,
> interaction models directly support methods to show that content is
> fresh (challenge-response & streamed remote attestation) or don't have
> to because the content shows that by itself (time-based remote
> attestation).
>
> Problem: Reiterating how the common models work and how they are
> related in every solution draft can prove to be error prone,
> ambiguous, difficult to compare in English, or simply be redundant.
>
> Solution: Describing them once and then referencing the bulk of it
> eliminates the need for text cloning, danger of inconsistencies/subtle
> deviations, and redundancy in general.
>
>
> The remaining question is: where should this content about interaction
> model go?
>
>
> At the last meeting four options were presented:
>
> * Option 1: standalone (one I-D for each model)
> * Option 2: standalone (one I-D for all models)
> * Option 3: all three models merged into the architecture I-D
> * Option 4: each model merged into a separate solution I-D
>
> Each option has pros and cons. So, I'd like to bring this question to
> the list (,finally... I can be quite slow).
>
> There is of course the current reference interaction model I-D that
> talks about the three models and how direct anonymous attestation can
> be enabled by all three of them. Please have a quick look, maybe that
> helps to provide some feedback here :) Five drafts reference this one
> at the moment:
>
> > https://datatracker.ietf.org/doc/draft-birkholz-rats-reference-interaction-model/
>
> We will dedicate some time to this topic on Tuesday and hope for some
> feedback!

Highlighting common interaction patterns that can be grasped without
delving into protocol details seems like a useful undertaking.  For
that reason I'd prefer we don't do 4.

Interaction models looks orthogonal to the architecture, so I wouldn't
conflate the two - risking slowing down delivery of the arch doc, as
already noted by Guy and Eric.  This rules out 3.

Which leaves me with either 1 or 2.  However, if I have to choose, I
wouldn't do 1 since it's not clear to me what the advantage of having
multiple separate docs is.  (I seem to only see disadvantages: editorial
and process overhead, scattered reading.)

Therefore, I'd say 2.

Cheers!




IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.