Re: Confused by recurring-duration

Ashok Malhotra/Watson/IBM wrote:
> 
> Oh boy!  I just looked at the draft and realized there is no discussion
> of a value space for recurringDuration.  Thank you, James, for catching
> this.  My apologies for the omission.  Paul, we need to fix this.
> 
> Let me take a stab at defining the value space for recurringDuration:
> "The value space for recurringDuration is a series of timeDurations
> that recur with some frequency.  The frequency of recurrence is defined
> as the timeDuration between two occurrences."

Each member of the value space is a series.  Thus the value space is
itself a set of series.  The members of each series in the value are
surely timePeriods not timeDurations (they start at a particular instant
of time, and persist for a particular duration).  Only a series that
satisfies the following constraints can be members of the value space:

- fixed duration: the duration of each member of the series must be the
same

- fixed period: the duration from the start of one member to the start
of another member must be the same for any two consecutive members of
the series

- infinite: every member has a preceding and a following member

In other words it is a single duration of time that recurs starting at a
series of equally-separated instants of times.

A single period of time is modelled as an infinite series with a zero
period.

The lexical space represents the value space by representing the start
of one the time periods that is a member of the series.  The choice of
which time period to represent is arbitrary.  Thus

<recurringDuration duration="P0Y1M"
period="P1Y">2000-01-01T00:00:00</recurringDuration>

represents the same value as

<recurringDuration duration="P0Y1M"
period="P1Y">2001-01-01T00:00:00</recurringDuration>

I find it hard to see how you can pick a single canonical lexical
representation.  Perhaps recurringDuration should be abstract.

James

Received on Friday, 15 September 2000 01:42:15 UTC