On 19 January 2011 15:57, C. M. Sperberg-McQueen <[log in to unmask]> wrote:
It depends, I guess, on whether you are seeking to
define a format for user input or for interchange among
parties.  It's often quite useful to accept a lot of forms
in user input that one does not want in a database or
document to be stored and interchanged.

Users may well type "19.01.2011" or "1/19/2011" or
"18 Jan 2011" or "on the 19th of this month", all meaning
the same thing.  But in metadata (as opposed to the
transcription of historical or legal documents), is there
an important goal to be achieved by preserving the differences
among those forms, given that (by hypothesis) they all
mean the same thing?
The potential requirement in my field is definitely the interchange of information, and not the user input. For e-portfolio systems, or indeed any system on which individuals may record their personal history, the interface will take care of the user input in whatever form is appropriate for that set of users. However, what would be really useful would be to be able to represent uncertain or approximate dates in information sent between different systems. Many people, particularly when recalling dates in the more distant past, cannot recall exactly when they happened, however they can either recall or work out some range within which the event must have happened. It would be useful to represent this in a machine-readable way.

As I have indicated before, the main thing that interests me here is how much clarity and simplicity we can get into a spec that is based on ISO 8601, taking away awkward things that are not used, and adding a general, clean and, to the extent that it is possible, intuitive representation of uncertainty. I don't think the draft is there yet, but it seems to be getting closer... :-)


Simon Grant
+44 7710031657