Re: [regext] Murray Kucherawy's Discuss on draft-ietf-regext-data-escrow-07: (with DISCUSS and COMMENT)

["Murray S. Kucherawy" <[e-mail]>]

On Thu, Apr 9, 2020 at 6:45 AM Barry Leiba <[e-mail]> wrote:

> > Section 5.1.3: "This element SHOULD be present in deposits of type
> Incremental
> > or Differential."  This makes it sound like a deposit of those two types
> not
> > using this element might be non-compliant.  I suggest instead "This
> element is
> > only used in Incremental and Differential deposits."  (Or instead of
> "used",
> > maybe "meaningful".)
> >
> > Section 5.1.4: " It SHOULD be present in all type of deposits."  Same
> issue.
> > Maybe "It is valid for use in all types of deposits."
>
> I don't understand this: "SHOULD" means that "there may exist valid
> reasons in particular circumstances to ignore a particular item, but
> the full implications must be understood and carefully weighed before
> choosing a different course."  The text in this document isn't saying
> that these make sense in the places they specify: the document *does*
> mean "SHOULD" here.  Certainly, "there are no relevant items to
> include" is a valid reason.
>
> What's the point of your objection?
>

The way I have always seen SHOULD used, consistent with that definition,
causes me to read 5.1.3 in a way that expects the "delete" element to be
present other than in exceptional circumstances.  However, I don't think
(and based on his comments I believe Robert agrees, though he can correct
me if not) that the absence of a "delete" element is really unusual; a diff
might genuinely -- perhaps even commonly -- only contain new items that
need to be applied to the backup with no delete actions.  The same goes for
5.1.4, where "contents" could legitimately be omitted merely because the
diff describes some canceled registrations that need to be removed.

RFC2119 also says "In particular, [2119 words] MUST only be used where it
is actually required for interoperation or to limit behavior which has
potential for causing harm (e.g., limiting retransmisssions)" and I don't
thnk that's the case here.

-MSK