BTW, in addition to what I said below. In MOWS I have reflected
exactly what we have agreed to on the call when discussing this metrics problem
and that is: SinceReset metrics has ResetAt but no Duration and Interval metrics
has Duration and that would be the criteria for differentiating SinceReset and
Interval. So I would say that MOWS needs to be fixed to match the
decision.
We should discuss this again as not all may remember the decision the same
way.
We may certainly key off the ResetAt and make it not appear on Interval
metrics, but it is strange to assume that Interval metrics cannot be
reset.
-----Original Message----- From: Sedukhin, Igor S
Sent: Tue 11/30/2004 1:05 AM To: Fred Maciel;
wsdm@lists.oasis-open.org Cc: Subject: RE: [wsdm] MUWS
parts 1 and 2 and MOWS comments
1. For SinceReset metric having ResetAt attribute why is
Duration attribute needed?
2. If Duration = CurrentTime - ResetAt,
then why do we need both at all?
-- Igor Sedukhin ..
(igor.sedukhin@ca.com) -- (631) 342-4325 .. 1 CA Plaza, Islandia, NY
11749
-----Original Message----- From: Fred Maciel [mailto:Fred.Maciel@hds.com] Sent:
Tuesday, November 30, 2004 12:41 AM To:
wsdm@lists.oasis-open.org Subject: [wsdm] MUWS parts 1 and 2 and MOWS
comments
Hi,
Here are Hitachi's
comments on MUWS and MOWS.
(1) There are a few of inconsistencies in
the metrics:
(1a) The MOWS spec specifies in lines 614-618 which
attributes must be reported for each kind of metric, which conflicts with
the MUWS spec (part 2, lines 836-849). For instance, for SinceReset
metrics, MOWS says that duration MUST NOT be reported and MUWS says that it
MUST. I presume that MUWS is the correct one (and where this text should
be) and my suggestion is to erase the paragraph in the MOWS
spec.
(1b) If I recall correctly, Heather said that we can discover the
kind of metric (Interval, PointInTime, SinceReset) from ResetAt and
Duration, and we decided to clarify it in the spec (again, part 2, lines
836-849). However, it's still not the case now: whether ResetAt must be
included or not for Interval and SinceReset is not specified, and Interval
and SinceReset can't be differentiated. Suggestion: ResetAt MUST be
included in SinceReset metrics, and MUST NOT be included in Interval ones.
But...
(1c) ... ResetAt in part 2, lines 787-792 is basically saying
that ResetAt is included if and only if the metric is resettable. Which
means that we cannot say that "ResetAt MUST be included in SinceReset
metrics" above for a SinceReset metric that's not resettable. We need to
discuss whether to adopt (1b) or (1c) -- I prefer (1b).
(2) We
are using http://schemas.xmlsoap.org/ws/2004/08/addressing
for WS-Addressing, but the June release of WSRF (which we are also
using) uses http://schemas.xmlsoap.org/ws/2003/03/addressing.
(3)
In MUWS part 2 it's not clear the difference between Duration
and CalculationInterval. Perhaps Duration is when the values were
collected to calculate the metric, and CalculationInterval is the period in
which the metric value is updated?
(4) Is there a clear definition
in MUWS on how to put the WSDM Event Format on a WSN Notification
Message?
(5) MUWS part 1, lines 468 and 470: these are elements, not
attributes.
** Mostly nits below **
(6) Part 2, line 872:
instead of "MUST include zero or more", how about "MAY include one or
more"? ("MUST include zero" is kinda strange...).
(7) Part 2, line 449
and 450: include this sentence in the previous one (it lacks a verb, to
begin with).
(8) Part 2, line 1042: Refrence -> Reference
(9)
Part 1, line 744: XX -> 12
Regards,
Fred Maciel.
To
unsubscribe from this mailing list (and be removed from the roster of the
OASIS TC), go to http://www.oasis-open.org/apps/org/workgroup/wsdm/members/leave_workgrou p.php.
To
unsubscribe from this mailing list (and be removed from the roster of the
OASIS TC), go to http://www.oasis-open.org/apps/org/workgroup/wsdm/members/leave_workgroup.php.
|