Jamie,
Respectfully, the KMIP TC co-chairs do not agree with your
characterization of events to date.
Regards,
On 12-Dec-17 5:47 PM, Jamie Clark
wrote:
Dear
members of the KMIP TC.
This
is a follow-up to our conversation at your TC meeting on
7 December.
OASIS staff asked you to postpone final action on your
2018 interop event until at least 14 December, so that
we can pass along the thoughts stated in this letter,
and gather your feedback on the rule issue described
below. We appreciate the TC's being willing to do
so. We plan to attend your meeting on 14
December to answer any further questions that the TC
members may have.
The
TC received a proposal from Tony Cox, on the day before
the last meeting, advocating that the TC decide not to
run an formal interop demo in 2018 [1],
because of OASIS' prohibition of ejections from the show
floor in the current version of our Interop Rules [2].
The draft names that single point of disagreement as the
reason to cancel a formal test for 2018.[3] The draft
at least inferentially leaves open the possibility of
other events for 2018.
As
we understood it, Mr. Cox's draft [4] proposed the
position that the KMIP Interop Demo event cannot be
conducted properly, without the ability to eject
registered participants at show time if they fail to
meet certain milestones,
with an exception for any members that the TC votes
specially to retain.[7]
At
your meeting on the 7th, we did not feel that we heard
many of the TC member companies express views on that
specific issue. So this message is designed to provide
background, and solicit your further input as TC
members.
To
summarize:
- As
you know, each TC is encouraged to adopt its own set
of "local rules" [5] to specify
test methods, timing, content, and
consequences. The KMIP TC needs to do so for the
2018 event soon, which drives this conversation's
timing,
- We
believe there has been good faith disagreement, for
several years, between our staff and the KMIP test
administrators, about whether a TC's "local rules"
should require ejection [6] of failing test
participants from the RSA show floor. Our 2017 policy
revision attempted to clarify this point.
- All
testing participants are disadvantaged, if individual
registrants repeatedly decline to participate fully or
timely -- with the results that the efforts of all
participants to fill the testing matrix may be forced
to limp along without success, causing all parties a
difficult technical rush at the end. So it's
reasonable if the TC wishes to adopt appropriate rules
to address that risk.
- However,
OASIS' staff view is that neither we, nor the TC, nor
its volunteer test administrators, ought to be in a
position of making discretionary subjective value
judgments about registrants, nor to make last minute
ejections on site.[7] Our preferred approach is to
publish and apply objective criteria, against which
any registrant can be measured well in advance of any
dispute on the show floor, and which are applied
equally to all registrants without discretionary
exceptions.
- We
have suggested several alternatives, described below,
for TC "local rules" to address any continuing
problematic failures, in ways that we believe are
consistent with OASIS policy.
The
TC may wish to consider imposing rules, at the start of
the testing cycle (that is, now, for the 2018 event),
that state clearly that a registrant will:
- not
be admitted to the event demo, based on determinations
made well in advance of the event, for which all
registrants have ample notice, or
- suffer
a reduced or differently-captioned presence, or
- enjoy
different permissions or lead access at the test
event, so that their status is not misrepresented by
their presence,
in
each case based on specific failures to meet stated
objective milestones at certain times. OASIS staff
believe that such cases are rare, but could be handled
appropriately by objective approaches of that kind, if
they arise. Our policy does state that staff still
retains the right to review any such proposed rules for
fairness. But rule-based objective approaches generally
are easier to administer fairly, and also are less
likely to generate last-minute disputes at the physical
show floor event.
For
example:
- The
TC might decide that, if any test applicant has
participated X times in past years, and failed to meet
certain (or all) of the test milestones in each of
those X years, then they may be deemed ineligible to
participate in the next year's event, from the start.
- Also,
the TC might decide that any registered parties who
fail (using objective criteria specified in advance)
to meet X milestones during the year's test protocol
must be located physically separately, with different
signage, than other registrants.
- Obviously,
a rule that relies on past year performance would not
apply to new entrants who have not participated in
prior years.
We believe that rules of that type could appropriately be
inserted into your TC's "local rule" resolution for the
2018 event. If the TC feels that these are desirable
steps, but that the change would provide insufficient
notice to registrants for the April 2018 event, the TC
also could elect to make the changes effective for 2019.
OASIS
has no views on whether such rules to address difficult
cases are likely to be needed, in your TC. However, we
believe we should support the TC's preference, so long
as it is based on objective criteria, measured with
specificity and in advance.
The
foregoing options directly address question raised by
two members at your December 7 meeting about what other
options, than ejection from the floor, may be available
to address cases of continued problematic failure. It
remains for the TC, through each of its members, to
assess our suggestions, and decide how the TC would like
to proceed with a testing plan, and with what
consequences for test failure, in light of our current
policy.
Please
note, the consortium-level Interop Demo Policy is a
creation of our staff. We're human, so we also must be
open to feedback suggesting that our policy choices are
flawed. If a significant independent number of TC
members believe that our view on last-minute ejection is
wrong, we want to hear that, and I expect our
Board does as well. We invite participating companies
and experts to discuss these issues further within the
TC, but also welcome anyone who wishes to communicate
their views to us directly, either through our staff or
our Board chair, Gershon Janssen, who is copied here.
Thank
you for your attention, and bearing through this long
communication. Ultimately, all such events and programs
are voluntary, like open standards themselves, and rely on
the good will of the participants. We welcome your
feedback both negative and positive on how best to
administer them.
Cordially
Jamie
FOOTNOTE
REFERENCES
[1]
Mr, Cox's 7 Dec proposal to the TC can be found here: https://lists.oasis-open.org/archives/kmip/201712/msg00005.html
From the 8th paragraph of that draft: "KMIP TC finds itself
unable to endorse or support an Interop Demonstration[.]"
[2]
The OASIS consortium-level policy can be found at: https://www.oasis-open.org/policies-guidelines/interop.
Its Section 7.2 was revised in February 2017 to state that a
member who simply fails the test matrix "retains the option to
showcase its product." This is the policy position with which
Mr. Cox's letter disagrees.
[3]
From Mr, Cox's 7 Dec draft, 2nd paragraph: "However, we have
now reached an impasse over one particular issue."
[4] Mr, Cox's 7 Dec draft, in its 3rd paragraph, says:
"OASIS Staff firmly believe that a member shall never be able
to be precluded from participation in an Interop Demonstration
even if the technical or behaviour requirements have not been
met." OASIS staff does not see our position the same way --
please see footnote [6].
[5] The TC's adopted local rules" for last year's event,
dated October 2016 for the February 2017 RSA conference,
called the "KMIP Formal Interop Process v1.4a," which can be
found here: https://www.oasis-open.org/committees/download.php/59127/KMIP%20Formal%20Interop%20Process-V1.4a%20(Oct2016).doc
[6] Please note, OASIS staff does not rule out
the possibility that extraordinary bad behavior might
result in staff ejecting a participant. Imagine, for example,
someone who started a fight on the show floor, or sabotaged
some other member's device. The point of our policy is that mere
failure to pass the test, measured immediately before
the show opens and subject to a discretionary free pass from
the TC, is an insufficiently-objective and
poorly-timed.approach.
[7] Section 4.1.1 of the KMIP 2016 local rules would eject
all registrants who fail the interop tests, but with a
special exception for any that the TC and the test group
choose to invite back to the show floor by a special
majority vote. OASIS staff sees that as a subjective and thus
problematic approach. We acknowledge that the TC enacted such
a rule in prior years. Staff simply should have noticed and
raised that issue earlier.
|