W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > July to September 2005

Re: insufficient value range for duration and date types

From: Dave Peterson <davep@iit.edu>
Date: Tue, 30 Aug 2005 21:29:16 -0400
Message-Id: <a06230905bf3ab4cead7d@[]>
To: Thomas Fischer <Th.-Fischer@web.de>, www-xml-schema-comments@w3.org

At 2:22 PM -0600 050830, Thomas Fischer wrote:
>I can't find any discussion about insufficient value range
>for the types date, dateTime, gYear, gYearMonth and
>duration, but I think  there is a gap: Infinity
>I have an element "event" with the attributes dateTime,
>duration and interval If I like to define the event as
>starting at a point and continue for ever, i can't realize
>this with the given nifty event element and no XSLT2/EXSLT
>date function can handle infinities.
>When there are infinite durations (and that's the real
>life) it's also necessary to have infinite values for
>the types date, dateTime, gYear, gYearMonth
>  date + infinite duration = infinite date

Real life it may be, but it does not correspond to our datatype.
So this would clearly be an extension (like adding plus and minus
infinities to the real numbers, or adding infinities in all Argande-
plane directions--and maybe an undirected infinity as well--to the
complex numbers: you clearly get a new datatype).  Both versions
of the datatype would have their uses.  I wonder if this addition
would pass the "80-20 test".

I'm curious as to what real life anchored durations you have in
mind that are half-infinite intervals.  Do they occur in situations
when you must also deal with finite anchored durations?  If not,
then you can just model them by using their anchor point-in-time.
Dave Peterson

Received on Wednesday, 31 August 2005 01:28:57 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:50:03 UTC