On Tue, 12 Jul 2011 10:51:40 -0400, Ray Denenberg, Library of Congress wrote
> If we are going to define an interval to mean one or the other
> (1) that interval of time, or
> (2) when an event occurred, e.g. "it occurred sometime between 1964 and
> The we need to go with (1), because that's how ISO 8601 defines it.
ISO 8601 does not to my eyes define the application of interval only its
> (Aside: although that itself isn't completely clear, but is "best
interpretation. And, it isn't clear that ISO 8601 has the same concept of
precision as we seem to be adopting; 8601 would see 1964/2008 as meaning
the interval starting at the very beginning of 1964 and ending at the very
end of 2008,
I do not agree. If this was the intent (e.g. 1964 as shorthand of 1964-01-
01T00:00.000Z) then it has failed as I do not think I have ever seen an
application that adheres to that.
> while we see it as the interval beginning sometime in 1964 and ending
sometime in 2008. However, 8601 isn't even completely clear on that point,
and I have no problem with our view.)
I think its highly mislead to use the term "sometime". Would one use it to
describe an event whose date value is "1964-06-08T12:33Z"? Did the event
really occur within the first (or last second) of the minute? Again.. we are
back to p-r-e-c-i-s-i-o-n... ! :)
> But look at it this way: suppose you want to say "the event occurred at
some intant between 1964 and 2008". The data element that you are saying
this for could have a definition like "the interval within which the event
> So as you say "you are not allowing people to unambiguously state whether
they are referring to an interval or a point in time within that interval".
Is that really necessary? When you state a dateTime expression, there has to
be some context, doesn't there? In other words, you are stating it as the
value of a data element, right? And the data elements definition would
dictate which interpretation applies.
> From: Discussion of the Developing Date/Time Standards
[mailto:[log in to unmask]] On Behalf Of Simon Grant
> Sent: Tuesday, July 12, 2011 2:50 AM
> To: [log in to unmask]
> Subject: Re: [DATETIME] interval precision
> Ray, I think you've hit on a real point there.
> On 11 July 2011 22:48, Denenberg, Ray <[log in to unmask]> wrote:
> (1) Does it mean that interval of time, or
> (2) Is it a way to represent when an event occurred, saying "it occurred
sometime between 1964 and 2008".
> And my understanding is that this question is out of scope, it could be
either, it is a matter for the application to specify for a given data
> Personally, I cannot see any advantage in this ambiguity. It would be
really helpful if, at least, it was completely clear from the syntax whether
one was referring to the whole of an interval, or to a moment in time
somewhere in an interval.
> Think of it like this - the way the proposal is at the moment, you are not
allowing people to unambiguously state whether they are referring to an
interval or a point in time within that interval.
> And it would be perfectly possible - quite easy in fact - with appropriate
modifications to the spec.
> Simon Grant
> +44 7710031657
Edward C. Zimmermann, NONMONOTONIC LAB