Ok take this concept of someone's wedding date, and how it fits with this specification. It is completely outside, though complementary, to the spec. That is, there is an application in the model that makes use of the specification.
That application may be a process, for example one that creates metadata about an individual. So you might have a data dictionary that looks something like
Name mandatory, nonrepeatable type: string
Date of birth mandatory, nonrepeatable type: edtf
Country of origin mandatory, nonrepeatable type: ISO 3166
Country of residence optional, nonrepeatable type: ISO 3166
color of eyes optional, nonrepeatable oneOf[blue,brown, green ....]
hair color optional, nonrepeatable oneOf[black,brown, gray, blond ....]
wedding date optional, repeatable type: edtf
date of death optional, nonrepeatable type: edtf
The data dictionary is part of the application and completely outside the scope of the edtf spec. The role of the edtf spec in this process is to provide a format for those data types specified as 'edtf'.
The "wedding date" could have whatever "description" you want as part of the data dictionary and it is completely outside the scope of the datetime spec, and it is repeatable for those who have ventured more than once, and optional for those never married.
From: Discussion of the Developing Date/Time Standards [mailto:[log in to unmask]] On Behalf Of Simon Grant
Sent: Tuesday, December 14, 2010 8:07 AM
To: [log in to unmask]
Subject: Re: [DATETIME] precision
On 14 December 2010 10:44, Jakob Voss <[log in to unmask]> wrote:
Edward C. Zimmermann wrote:
On Thu, 9 Dec 2010 17:24:43 -0500, Denenberg, Ray wrote
is a decade: 10 years.
That's the basic fallacy. A decade "is" not the set of its years, but it "consists of" the set of its years. Similar a day is more than the set of its hours, it's just another kind of concept. For instance your wedding day is neither an interval, nor a set of moments. Sure you can mathematically define dates in such a manner, but this completely removes the human aspect of dates (call it cultural, sociological or anything, but *not* "purely logical"). Without this aspect you can just use a simple model based points, intervals, and precisions in time, all defined by seconds. Computers do not need minutes, hours, days, month, years etc. Human perceptions of dates are different.
This nicely highlights the need for conceptual clarity about what kinds of things we are talking about. However, the approach of stating what the "facts" are about what concepts mean is, I believe, mistaken. Concepts are useful inasmuch as people agree on their meaning enough to have meaningful conversations, make decisions that stick, etc. A process of consensus building is necessary to achieve this state of agreement, and that can be educative; pronouncements don't usually work.
"Your wedding day" is I guess a concept that we don't really want to be dealing with directly. When I say that my wedding was on a certain day, I mean that the actions that constituted that wedding took place within that day. Unfortunately the concept of someone's "wedding day" is ill-defined: people could use the phrase to mean the day on which their wedding took place, as in "were you born on my wedding day?", or it could be used for the set of events that constituted the wedding. We can have a set of photos categorised "my wedding day", and it would be rather strange to see random things from around the world that happened on that day, or photographs of a clock, for instance. Personally, I don't think of a decade in that way, but more just as a period of time.
Human perceptions of time-related things, and their concepts, are indeed different, and they can change and develop. I don't think that this list would be well employed trying to tease out all the ways in which people think of time and events in time. I do think it is potentially useful to give people a well-defined and richer vocabulary for expressing when things happened.