The reason why unknown values should be allowed is that it makes the distinction between uncertain and unset values.

The notation "?/?" carries extra meaning in that it specifies we are dealing with an interval and not a date, time, or date/time; however, I would question whether or not the distinction needs to be made.

Hypothetically, we could have a system of describing unknown values which retains the type of time entity that is represented. For example:

"?": Unknown date.
"T?": Unknown time.
"?T?": Unknown date/time.
"P?": Unknown duration.
"?/?": Unknown interval.
"R?/?: Unknown repeating interval.

But when is it important to specify the type of time entity?

Maybe using "?" for all time entities would suffice, and the type of entity it represents could be determined by the context?

Nathan

On Sat, Aug 1, 2015 at 2:58 PM, Saašha Metsärantala wrote:
Hello!

When we were working with EDTF, the question "When was this text written?" could be considered a typical use-case. Now, with ISO-8601-2, this question can (of course) be considered a use-case too, but maybe not a really TYPICAL use-case. I consider that our focus shift from EDTF to ISO-8601-2 requires a widening our perspectives.

Being acquainted with Gauss' and Cantor's theories (and the continuum hypothesis and ZFC etc.) does not mean that we need to limit ourselves to these. Of course they can be used as a basis to build some use-cases, but fiction, the future (and of course future fiction) (will) offer other perspectives. Indeed, the past has already offered other perspectives.

The librarian approach is a great starting point,
Yes, it is really! It also offers many good use-cases!

but we sure don't want to be limited by it. Kronecker notwithstanding, mathematicians never stopped with just the integers.
Not even with the octonions, but their "norms" get damaged ... Well, let's leave those (nearly off-topic) algebras and let's focus on ISO-8601-2, again.

I hope you don't misunderstand me. I do not mean that we HAVE to cover everything. But I consider that there is a need to clarify (both for ourselves and) for implementers and other readers of the ISO-8601-2 specifications what we (try to) cover and our awareness of what ISO-8601-2 doesn't cover. My conclusion is that:

- If we exclude "?/?" (and the like) AND clarify WHY we choose to exclude (some of) them, it will be OK for me.

- If we include "?/?" (and the like) AND clarify WHY we choose to include (some of) them, it will be OK for me.

- I discourage an unmotivated exclusion (or inclusion) of "?/?" (and the like). I consider that we really need to continue to clarify our awareness of the width of the field in which we have come when coming into ISO.

My suggestion is to continue to clarify those things before we decide whether we will include or exclude "?/?" (and the like).

Regards!

Saašha,