Re: [Anima] Roman Danyliw's No Objection on draft-ietf-anima-grasp-api-08: (with COMMENT)

[Brian E Carpenter <[e-mail]>]

Hi Roman,

Thanks. Assuming we get to "New ID needed" we'll
look at all these. One or two quick comments in line.

On 02-Dec-20 13:00, Roman Danyliw via Datatracker wrote:
> Roman Danyliw has entered the following ballot position for
> draft-ietf-anima-grasp-api-08: No Objection
> 
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
> 
> 
> Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
> for more information about IESG DISCUSS and COMMENT positions.
> 
> 
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-anima-grasp-api/
> 
> 
> 
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
> 
> Thank for responding to the SECDIR reviewer and thank you to Joseph Salowey for
> performing it.
> 
> ** Since this is an API spec a few more example pseudo code snippets showing
> common ASA “tasks” invoking this API from both sides of the connection (like
> Figure 2) would be very helpful.

We have some of that in draft-ietf-anima-asa-guidelines, which is
still evolving.

> ** More precise references to draft-ietf-anima-grasp might helpful to
> implementers (e.g., in Section 2.3.2.3, “… default GRASP_DEF_LOOPCT, see
> [I-D.ietf-anima-grasp]” ==> “... see Section 2.6 of [I-d.ietf-anima-grasp]”)

Assuming the GRASP spec soon enters AUTH48, we can actually get the
references right.

> 
> ** Section 1.  Per “An ASA runs in an ACP node and therefore inherits all its
> security properties, i.e., message integrity, message confidentiality and the
> fact that unauthorized nodes cannot join the ACP.”, in the spirit of precise,
> things like message integrity and message confidentiality are not properties of
> the ASA or of the ACP _node_ but instead properties of the protocol used on the
> control plane.
> 
> ** Section 2.1.  Recommend using consistent terminology.  In this section ASA
> call a “GRASP module”.  However, Section 1 lays out an architecture of GRASP
> Core + API.
> 
> ** Section 2.2.  I found the placement of this section confusing.  There is a
> discussion of the calling conventions for an API that hasn’t been discussed
> yet.  IMO, this should be after Section 2.3.  That said, thanks for describing
> these different calling conventions.  Showing these in examples would be very
> helpful.
> 
> ** Section 2.2.2.2.  Per the definition of TTL, is it worth clarifying here and
> in the subsequent descriptions that this is an unsigned of a particular size
> (unsigned 32-bit at least) per Section 5 of draft-ietf-anima-grasp?
> 
> ** Section 2.3.2.3.  Is it worth clarifying that loop_count should be between 0
> and 255 per Section 5 of the draft-ietf-anima-grasp?
> 
> ** Section 2.3.2.3.  Provide a normative reference to which version of C and
> Python will be used.
> 
> ** Section 2.3.2.3.  If an older C is used, is “char *name” the right way to
> handle a UTF-8 string?

Dunno. Has there ever been a stable answer to "How do I process strings in C?"
 
> ** Section 2.3.2.3. Per the C data structure of an objective, should loop_count
> and value_size be unsigned integers of some kind?
> 
> ** Section 2.3.2.3.  Why does the Python implementation set a default value of
> loop_count but C does not?
> 
> ** Section 2.3.2.3.  Please provide a reference to libcbor
> 
> ** These examples in C and Python found Section 2.3.2.3 were helpful.  I was
> hoping to find them in the other sections.  Also a C-style .h file with
> function prototypes and constants would also be nice (e.g., GRASP_DEF_TIMEOUT,
> IPPROTO_*, all the error types)
> 
> ** Section 2.3.4.  Typo. s/tiemout/timeout/
> 
> ** Section 2.3.2.4.  The constants IPPROTO_TCP and IPPROTO_UDP aren’t defined
> here.  Recommend a reference to the grasp draft.

Yes, I guess that's correct, although the actual values are IANA-assigned
and widely used in the socket API.

> 
> ** Section 2.3.7.  Double checking -- per the info input parameter, is the ASA
> supposed to provide this content or is this something from GRASP Core?
> 
> ** Appendix A.  This list doesn’t appear to be a complete crosswalk of function
> to error codes to possible APIs.  For example, “NotObj” is listed as a general
> error code, but would that get returned by register_asa()?
> 
> ** Per the GENART Review, IMO, Paul makes a number of good points, in
> particular: -- a reference or further explanation of the flow for dry run and
> how this would be used in other API calls
> 
> -- additional clarifying language on request_negotiate
> 
> -- Renaming the “session nonce” to “session handle” (or something like it)
> might improve clarity so the API doesn’t have to deal with multiple “nonce”
> 
> 
> 
>