Re-wording of issue 27 consensus (Re: [Ietf-calsify] Resolutions of issues 8, 68 and 69: DST discontinuities)

[Jonathan Lennox]

Following today's jabber discussion, I'd also suggest clarifing issue 27's
consensus, in the new terminology, to say that DTEND - DTSTART is always an
accurate duration, not a nominal duration.  Thus, issue 27's example 2

  DTSTART;TZID=America/Montreal:20070311T000000
  DTEND;TZID=America/Montreal:20070312T000000
  RRULE:FREQ=DAILY;COUNT=2

is exactly equivalent to

  DTSTART;TZID=America/Montreal:20070311T000000
  DURATION:PT23H
  RRULE:FREQ=DAILY;COUNT=2

because 2007-03-12 00:00:00 EDT is 23 elapsed hours after 2007-03-11
00:00:00 EST in America/Montreal.

I'm pretty sure this is what the consensus on issue 27 was, but I think this
makes things clearer and is easier to understand.

Does this seem reasonable?

On Thursday, March 1 2007, "Jonathan Lennox" wrote to "<ietf-calsify at osafoundation.org>" saying:

> For issue 8 (P1D vs. PT24H):
> 
> ISO 8601 defines D and W durations (and also larger ones that iCalendar
> doesn't use) as being "nominal" durations, while the smaller periods are
> "accurate" durations".  (See the jabber logs for the full quote from 8601).
> 
> To align with 8601 and remove ambiguity, then, we propose:
> * D and W durations are nominal, i.e. refer to the same wall-clock time n
> days or weeks later.
> * H, M, and D durations are accurate, i.e. refer to an actual number of
> elapsed hours, minutes, or seconds.
> * We forbid DURATION values that mix nominal durations (W [already
> forbidden] and D) with accurate durations (H, M, and S).

-- 
Jonathan Lennox
lennox at cs dot columbia dot edu