RE: XSD 1.1 issue: when did timezones cease being durations?

On Tue, 2005-03-01 at 14:34, Dave Peterson wrote:

> 
> If that means you've stored one integer that is a multiple of 60 between
> -50400 and 50400, rather than storing one integer between -840 and 840,
> then we're quite close--I don't see that it matters much.  But a
> dayTimeDuration value is not an integer, it's a pair of integers.  Is
> that really what you want to store?

Who's talking about storage?  We're not specifying
an implementation, or even a design for an implementation.
We're writing a specification.  What someone stores is
not our concern, and there is no benefit to the spec
in attempting to 'optimize' storage by treating
timezone information as an integer instead of as
a duration.

You mention a bug found earlier in the idea that
timezone information is a duration; at the risk
of replowing well known ground, can you remind 
the rest of us of what it was?

Michael