Not sure whether we need a meeting today...other than announcements.
Toby may not be available.
I spent some quality time with Mike yesterday, but I'm not ready yet
for PIM WD08; I'm working to clarify the issues that the example
work-through and the time with Mike raised.
I discovered some issues in working through examples from CS01, and
the new internet drafts on Icalendar relationships
(
http://tools.ietf.org/html/draft-douglass-ical-relations-02 ) and
availability (
http://tools.ietf.org/search/draft-daboo-calendar-availability-05
).
Key issues:
- Some of the CS01 examples are a bit obscure but apparently
correct; some I'm not sure of yet. So the examples are of
varying quality IMO but non-normative.
- The terminology for granularity and precision is not as clear
as it could be, and suggests different PIM approaches
- granularity appears only in vavailability. Gluons do not have
granularity, which is not perfectly aligned with the CS01 text.
- precision suggests a global configuration or capability for
resolution on times but is only in ToleranceValueType (as in
CS01) - this needs consideration as to the scope and inheritance
- Relationship of granularity and precision could be more clear
(and the clarity will guide the PIM
- The division of iCal/xCal "relationships" into
"relationshipType" and "temporalRelationshipType" needs to be
re-examined. These are combined in the relations draft
- in xCal tZ is a property that can apply to any object; in PIM
WD07 it's an attibute of IntervalType and its subclass
GluonType, which has dtStart (and in PSMs could have dtEnd);
inheritance needs to be clear. Is it an issue to have tZ apply
to an entire sequence? I would suggest that a single timezone is
entirely appropriate and a simplification.
- Per meeting discussion, tolerance is part of
DurationValueType, but the duration is perhaps a degree of
freedom too much if you also add fuzziness (startafter,
startbefore, endafter, endbefore) to the end points. This may be
better addressed as a constraint on values.
- Diagrams of examples (object diagrams in UML) aren't as clear
as they could be because the tool (Enterprise Architect) doesn't
have a good way of expressing nested object instances. I've
attached a first pass at Examples 3-12 - there are also no clean
ways to draw arrows to show actual time sequences. I'm planning
separate timeline drawings.
I'm hoping for a full update with all the examples circa February
24, as I am tied up with the Innovative Smart Grid Technologies
conference where Toby and I have a paper (unrelated to scheduling,
but related to microgrids and structured energy).
Thanks!
bill