[Css-csts] Forward Frame CSTS WG Review Package on CWE

[John Pietras]

CSTSWG colleagues ---
I have been informed that the FF-CSTS prototype testing has successfully concluded with no changes required of the FF-CSTS specification. I have therefore put together a working group review package and posted it to the CWE at URL:
https://cwe.ccsds.org/css/docs/CSS-CSTS/CWE%20Private/Future%20Services%20using%20Toolkit/Forward%20Frame%20CSTS/FF-CSTS-922x3r2.2-1200303-CSTSWG_Review.zip

The zip file contains 3 files:

  1.  An annotated version of the FF-CSTS book, in which the annotation consists of Word markup balloons that identify the reasons for the changes from the internationally-reviewed Red Book version. Most of these references are to the RIDs that caused them, but a few others refer to changes resulting in test compilations if the ASN.1 modules and typo that were discovered along the way. When the book is ready to go to the Secretariat for publication, these annotations will of course be removed.
  2.  An updated copy of the RID resolutions. Note at all RIDs and pseudo RIDs have been resolved, accepted by the WG, and incorporated into the document except the following three, which all stem from Peter Shames’ request to have the book more explicitly address the role of the FF service in Delay-Tolerant Networking (DTN) and Pater’s subsequent review of the document as a whole:

  1.  Pseudo RID JPL PS-1: This pseudo RID documents Peter Shames’ request for an explicit discussion of the role of the FF service in Delay-Tolerant Networking (DTN) configurations. Proposed changes have already been drafted in the book. Those changes have been accepted by the pseudo RID author. If and when the CSTSWG agrees with the changes, pseudo RID JPL-PS-1 will be closed.
  2.  Pseudo RID JPL PS-2: This pseudo RID documents multiple comments made by Peter Shames on the Red Book. Proposed resolutions for all items in the pseudo RID have been generated, either as changes that have already been drafted in the book or as explanations of questions asked. All of the proposed resolutions have been accepted by the pseudo RID author. If and when the CSTSWG agrees with the changes, pseudo RID JPL-PS-2 will be closed.
  3.  Pseudo RID JPL PS-3: This pseudo RID deals with the change in the CCSDS Blue Book “boiler plate” introduction to the References sections of CSTS specifications. For reasons having to do with the close synchronization of CSTS specification versions and CSTS SFW versions, the introductory paragraph now prohibits subsequent versions of the SFW from being used with any given issue of a CSTS specification. This change in approach has elicited concern by Peter about the desirability and feasibility of essentially maintaining multiple active versions of the SFW book. I have copied the relevant portions of our email exchange at the bottom of this email (below my signature). The most recent response (today) from Peter is “I am willing to close out the comment, but only if the CSS CSTS WG makes a commitment to a) making sure that this situation] is made blazingly clear, and b) make a commitment to fixing this at the soonest possible opportunity.“

Pseudo RIDs JLP-PS-1 and JPL-PS-2 only require the WG’s concurrence for their closer. However, JP-PS-3 requires some thought on how to clarify the situation and how the WG will prioritize updating the MD and TD books.

I would like to request that we put pseudo-RID JPL-PS-3 on the agenda for next Thursday’s telecon. I have included the relevant contents of the emails between Peter and myself in the text of pseudo-RID JPL-PS-3. I would encourage WG members to read through that pseudo RID before next week’s telecon so as many as possible of us will have some understanding of the situation.

Thanks.

Best regards,
John

EMAIL EXCHANGE BWTWEEN PETER SHAMES AND JOHN PIETRAS

Original Comment from Peter: Section 1.9 (References): Regarding the sentence “All other publications in the list are subject to revision, and users of this document are encouraged to investigate the possibility of applying the most recent editions of the publications (other than the Specification Framework) indicated below.”  RID author commented “Is it really true that ALL of these other references are suspect?  Even the most recent of them?”
Response (26 November 2019)  This is a variation of the standard boilerplate for all Blue Books, which is ”All publications are subject to revision, and users of this document are encouraged to investigate the possibility of applying the most recent editions of the publications indicated below” (CCSDS A20.0-Y-4 Cor. 1, February 2015, 3.4.1.8 (a) (2)). Our CSTS modification (which will apply to all CSTS specifications) calls out the fact that a given issue of a CSTS spec is tied to a single issue of the Framework and so no version of the Framework beyond the one explicitly cited can possibly be applicable to a particular issue of the CSTS spec.
Reply from Peter (4 December 2019)  For sec 1.9 it was what I read as a peculiar shift of language that bothered me. It is always that case the specs are only aligned with those others that are current at the time of publication. This is not any different for the Framework than for any other doc. And it is entirely possible that a future change may be made to the Framework without necessarily changing all of the other docs that use it. I think what is different here is that this doc is explicitly tied to the version of the Framework that contains the updates that are require to support FF-CSTS. I think you can just state that somewhere in the doc.
Pietras esponse to reply (24 February 2020)  In response to the statement “I think what is different here is that this doc is explicitly tied to the version of the Framework that contains the updates that are required to support FF-CSTS.” That statement is not exactly true, because although future versions of the Framework could theoretically also be used by “this” version of the FF spec insofar as the functionality defined therein doesn’t change for the procedures that are used by FF, there are version-specific numbers that are embedded in the SFW ASN.1 modules that increment with every version of the SFW and the FF procedures are tied to the specific SFW version numbers. A CSTS will work with *only* the version of the SFW that is specified in the Reference Documents section. And that’s just not FF-CSTS: it’ll be true for every CSTS. Plans for the next revision of the CSTS Guidelines are to include that re-wording of the boilerplate for all CSTS specifications.
So it’s not that this is a one-off situation for FF-CSTS, it’s something that has to be addressed for all CSTS specifications. The WG chose to address it right up front, and to explicitly clarify that the boilerplate statement “users of this document are encouraged to investigate the possibility of applying the most recent editions of the publications indicated below” does not, and cannot, apply to the SFW. The feeling of the WG was that if this was left “unchallenged” in the Reference Documents boilerplate the constraint might still be overlooked even if it were stated somewhere else (later) in the document.

Reply from Peter (2 March 2020): What I am getting from this is that the WG has adopted an SFW and associated standard versioning approach, if I understand it correctly, such that each CSTS spec is tied to a specific version of the SFW.  What this says to me is that there is no stated policy for backwards compatibility such that any future changes to the SFW would continue to be compatible with earlier CSTS specs.  If that is truly the case, and I can understand situations where future changes might break that "rule", the consequence would appear to be that the CSS (and the users of these specs) will have to maintain ALL versions of the SFW forever.   This seems really cumbersome at best. Am I misunderstanding this?

Pietras Response to Reply (3 March 2020): I think that you’ve got it right. However, there are a couple of additional points that I should have made:

  1.  Any new CSTS specification will always be written citing the then-current version of the SFW. There will be no option for, say, using the 2-issues-back version of the SFW as the basis of a CSTS.
  2.  The WG’s intention is to update all CSTS specifications to use the latest SFW ASAP (where the “soon” part is a function of available resources and other commitments by the WG). So older version of the SFW would be “active” only until all CSTS specs are aligned with the “current” SFW, at which point all the older books are truly “silverized”. Note that, like today, people may still choose to implement a silver version of a specification anyway (e.g., for backward compatibility with the user community) and this approach doesn’t address this issue any better or worse than it’s addressed for other CCSDS Silver books.



Case in point – the forthcoming FF-CSTS book uses “features” of the forthcoming SFW-B-2 (and indeed it was the need for these features that triggered many/most of the changes in SFW-B-2). But SFW-B-2 also changes (and simplifies) some of the data structures used not only by FF but also the already-blue MD and the on-the-verge-of-blue TD services. MD and TD can be changed from the B-1 to B-2 SFW data structures, and the WG intends to create projects to do so, but that will take time/resources to modify the respective “older” CSTS specs. The already-published MD book uses the SFW-B-1 data structures so it must reference SFW-B-1 or the implementation would be broken. In the case of the not-yet-published TD book (which also uses the SFW-B-1 data structures, the WG decided to go ahead and publish it with reference to SFW-B-1 in order to get it out the door, even though we know that it will have to be changed in the (near) future (I think the idea is to update the MD and TD books concurrently).



  1.  This may now seem like an administrative headache, but hopefully the changes to core elements of the SFW will settle down in the near future, and new versions of the SFW will only add new features, procedures, and/or operations that don’t affect existing CSTSes. In that (rosy) scenario, “updating” a CSTS spec to a new SFW will only involve a short Technical Corrigendum that points to the latest SFW issue and relevant paragraphs within that SFW issue.



Maybe one way to think about this is that, unlike the other references that CCSDS “encourages [users] to investigate the possibility of applying the most recent editions”, the CSTSWG *knows* what SFW version will work with the service as specified in the specification and essentially tells users up front not to waste their time thinking about it with respect to the SFW.

Peter’s Response (3 March): I am willing to close out the comment, but only if the CSS CSTS WG makes a commitment to a) making sure that this situation is made blazingly clear, and b) make a commitment to fixing this at the soonest possible opportunity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ccsds.org/pipermail/css-csts/attachments/20200303/97aecb2b/attachment-0001.htm>