Re: I've implemented the changes to xsd:dateTimeStamp

Regarding the new treatment of date-times:

In going through the diffs provided, I noticed a minor bug in the new 
text. In section 4.1
of syntax, the first sentence of the third para now reads:
"The datatypes owl:real and owl:decimal are defined as follows. "
While I believe it should read:
"The datatypes owl:real and owl:rational are defined as follows. "

I have more serious reservations about the addition of xsd:dateTime and 
perhaps
even the new interpretation of xsd:dateTimeStamp. We previously had 
quite constrained
but clear support for date-time. Now I am not sure what we are buying 
into by using the
xsd 1.1 part2 interpretation, so I would like a bit more time to look at 
the changes and at
xsd 1.1 to better understand this. One obvious concern is the partial 
ordering specified
in the current version of xsd for xsd:dateTimes without timezone offsets 
and those with.
Are we buying into this bit? I hope not.

-Evan

Boris Motik wrote:
> Hello,
>
> I have just implemented the change to the semantics of xsd:dateTimeStamp. While
> doing so, I took the liberty to clean up the datatype section a bit. The main
> problem was that the section was written as if it were defining various
> datatypes; for example, it said "The facets of xsd:string are such-and-such".
> This is clearly wrong: the facets of xsd:string are as defined in the XML Schema
> document, and there is nothing we can do to change this. Another problem was
> with the tables with the semantics of facets which were superfluous.
>
> To clarify all this, I have explicitly stated now that the sections defines only
> owl:real and owl:rational, and that the specification merely reuses the
> definitions of various datatypes. Hence, I've removed any attempts to (re)define
> XML Schema datatypes and have just added a bunch of examples by means of which
> we discuss certain consequences of their definition.
>
>
>   
...
> Finally, I would just like to point out that we should include xsd:dateTime into
> our datatype map now. There is no good technical reason not to do so;
> furthermore, most of the ontologies out there use xsd:dateTime rather than
> xsd:dateTimeStamp so, according to our current solution, such ontologies are not
> OWL 2 ontologies. I hope we can discuss and resolve this at the next teleconf.


/

/

Received on Monday, 30 March 2009 17:27:19 UTC