- From: Graham Gilmore <grahamg@steltor.com>
- Date: Fri, 22 Jun 2001 14:27:15 -0400
- To: "Libby Miller" <Libby.Miller@bristol.ac.uk>, "Alan Davies" <aland@steltor.com>, "RDF Calendar" <www-rdf-calendar@w3.org>
> I think I'm beginning to see a way through. My strong feeling is that > date-times need to be typed objects, and not just strings. This is > backed up by the DAML version of xml datatypes in RDF by Dan C. > However, unlike iCalendar rfc 2445, xml datatypes use UTC offsets for > timezones. Maybe what we could do is use objects for date-times but > allow UTC offsets to conform to xml datatypes. This would mean a > difference between icalendar in RDF and icalendar RFC 2445 though; and > I'm sure there are good reasons for the decision not to allow UTC offset > as part of the string (although it does appear as a property of a > date-time...). I've yet to finish going through the Calsch mailing list > archives - please forgive my ignorance. > > So this may seem like the worst of both worlds, since you get an object > and not a simple string, but without the precision of a timezone > definition object. But for many people it would be sufficient and it > conforms with standard RDF practice and xml datatypes. > > what do you reckon? Looking at a date-time value by itself, it doesn't need a full timezone specification; it only needs an offset (after all, one individual time instant cannot span a DST change). The reason we use the fuller concept of references to timezone objects in iCalendar is that "a time isn't always just one time". In iCalendar, the DTSTART is used to specify the beginning of a RRULE. However, the RRULE doesn't need to fully qualify all succeeding instances; for example, an RRULE may only say something like "repeat every week", as in: RRULE:FREQ=WEEKLY Since the RRULE doesn't mention anything about the other details of the succeeding recurrences, these elements are "inherited" from the DTSTART. The elements include, to use the WEEKLY example above, such things as which day of the week each recurrence is on, and (the core of the issue here discussed) the start time. As the time component (among other components) of the DTSTART is applied not just to that one instance of the event, but can potentially be applied to ALL of them, it is not sufficient to simply give an offset from UTC for the DTSTART. Some might argue that this overloading of the date-time type isn't really called for and there are better ways of expressing it. Probably you could find some means of communicating the intent of iCalendar while keeping the date-time type restricted to "only" a UTC offset, rather than some timezone entity reference (said reference could, for example, be added to the rdf version of RRULE instead, although you'd have to be careful if you wanted to maintain the implied iCalendar restriction that all the recurrences (including the one specified by the DTSTART) are in the same timezone). Graham
Received on Friday, 22 June 2001 14:29:32 UTC