[tei-council] Dates, Ranges, Periods and Durations

[Syd Bauman]

I have argued that (ISO 8601) durations should be permitted as the
value of value= of both <date> and <time> for a long time now, but up
until now I thought I was alone in this desire. This is why the
recommendatino in EDW90 for value= of <date> is
tei.data.temporalExpression, as opposed to 
    tei.data.temporalExpression | tei.data.duration
which might make more sense. However, as currently imagined,
tei.data.duration represents a W3C Schema duration, not an ISO 8601
duration. There are differences other than simple profiling.


> What I'm wondering about is whether we should be including the full
> scope of ISO 8601 including especially the use of ranges (or 'Time
> Intervals') and Durations.

My inclination is that yes, we should. But of course W3C only
supports a subset via their XML Schema (part II) datatypes (not to be
confused with the profile described for discussion puproses in the
W3C Note on the subject). This shouldn't stop us from supporting
other expressions, but should give us mild pause -- software that
supports XSD datatypes will be readily available; software that
supports other valid 8601 expressions may not be as easy to come by. 


> Maybe some of this is differences between ISO 8601:2004 and ISO
> 8601:2001?

I didn't know there was a 2004 version. Thanks for the info.


> I guess I was just wondering why we weren't seeming to support
> Durations, Periods, Ranges, in the datatype? But, knowing me, I've
> probably just misunderstood something.

I think it would be a fine idea. We would need to think pretty
seriously about where to follow 8601 and where to follow W3C. (E.g.,
IIRC, W3C does not permit a range indicated by a solidus between two
expressions, which 8601 does; W3C permits a + or - sign before a
duration, which 8601 does not.)