Re: representing scheduled events

Here's some info from the calsch group, I had assumed that
one of our more eager posters would have provided you with a
full response earlier...

iCalendar has a RECUR value type, which can express
quite complex (arguably too complex) recurrences
using units ranging from seconds to years (including days of
the week, month, year, etc.), for example "every Sunday in
January at 8:30 AM and 9:30 AM, every other year":

RRULE:FREQ=YEARLY;INTERVAL=2;BYMONTH=1;BYDAY=SU;BYHOUR=8,9;BYMINUTE=30

iCalendar also a PERIOD value type, used for durations.
For more information- http://ietf.org/rfc/rfc2445.txt 4.3.9 and
4.3.10

If you prefer XML syntax, there is a draft (which unfortunately
wasn't renewed, so has expired- I can send you a copy if you like)
that proposes an iCalendar DTD.

One thing to note is that a simplification of the recurrence rules
has been recently suggested, as another WG (iptel) wanted to use
the iCalendar syntax, but needed to be able to test whether a given
date intersected with a rule in a time linear with the length of the
rule. See http://ietf.org/internet-drafts/draft-ietf-iptel-cpl-04.txt
5.3 and Appendix A. Some posts related to this issue:
http://www.imc.org/ietf-calendar/mail-archive/msg06890.html
http://www.imc.org/ietf-calendar/mail-archive/msg06979.html

hope this is of some use...

--Alan


At 16:03 2001-04-17 -0400, Carl Benkley wrote:
>Mark,
>
>Your question may be quick, but I doubt there is any quick answer.  What
>you are looking for is a general way of expressing first occurence plus
>recurring interval (and possibly last occurence) My first guess would be
>to model it with the full range of possibilities that the data types in
>XML Schemas would allow.  A couple of valid possibilities within this
>context are "dateTime" plus "duration" (duration treated as a recurring
>interval), and "date" plus "gDay".  Unfortunately, the folks that
>brought us XML Schemas and the Gregorian ("g") date datatypes didn't
>give us a day-of-the-week, which seems critical to what you want to do.
>You could of course make up a dayOfWeek datatype.
>
>Also, I'm wondering if you are concerned only with events that happen in
>an instant of time.  Modeling events of non-zero duration obviously
>requires some way of modeling that duration.
>
>Carl Benkley
>Lead Information Systems Engineer
>MITRE Corporation
>
>
>Mark Nottingham wrote:
> >
> > Hello,
> >
> > I sent the attached message to the IETF calsch WG a few days ago, and
> > haven't gotten anything back. Does anyone here have strong feelings
> > about this, and/or a reference?
> >
> > I'd like to see something moving towards a (hopefully short) standard
> > here, as there are a number of different application domains. Or
> > perhaps I'm just missing one that already exists (crosses fingers)?
> >
> > Cheers,
> >
> > --
> > Mark Nottingham, Research Scientist
> > Akamai Technologies (San Mateo, CA USA)
> >
> >   ------------------------------------------------------------------------
> >
> > Subject: representing scheduled events
> > Date: Sun, 15 Apr 2001 14:01:49 -0700
> > From: Mark Nottingham <mnot@akamai.com>
> > To: ietf-calendar@imc.org
> >
> > Hello,
> >
> > Just a quick question. I've seen several fields where it's desireable
> > to model a scheduled event and then represent it in some form (XML or
> > a simple string, usually).
> >
> > For example, I'd want to model and express:
> >   - "Every Tueday at 6pm"
> >   - "The first Monday of the month at noon"
> >   - "Every other day at midnight"
> >
> > Possible uses include for HTTP cache freshness, recurring events in
> > RSS news feeds, and policy updates for things like P3P. In general,
> > I'm talking about representing a publication schedule.
> >
> > Does calsch have a solution for this, or any thoughts that would
> > point an effort to develop such a thing in the right direction?
> >
> > Cheers,
> >
> > --
> > Mark Nottingham, Research Scientist
> > Akamai Technologies (San Mateo, CA USA)

Received on Tuesday, 17 April 2001 17:06:25 UTC