- From: Michael Rys <mrys@microsoft.com>
- Date: Thu, 6 Nov 2003 09:08:14 -0800
- To: "james anderson" <james.anderson@setf.de>, "Jonathan Robie" <jonathan.robie@datadirect.com>
- Cc: "Kay, Michael" <Michael.Kay@softwareag.com>, <public-qt-comments@w3.org>
> thank you, but no, it does not.
>
> it neither deomstrates why one should need to do this nor why one
> should be allowed to. i keep asking for a demonstration because
neither
> an implementation of xslt, nor an implementation of xquery required
its
> implementation except specifically in order to satisfy the respective
> spec's stipulation that the function be defined. no robust application
> could seriously include the expression
>
> make-empty-element( xs:QName("james") )
>
> the whole point of this typing business is to make things robuster
> isn't it? why, then, specify your language to include operations which
> do not necessarily close over the data domains. the only thing which
> makes sense is something of the order of
>
> make-empty-element( xs:UName("", "james") )
>
> anything else is the logical equivalent of doing floating point
> arithmetic on pointers to known integer data.
>
> ...
[Michael Rys] I personally agree. Although, since there is no type
called xs:UName you would need a function.
Best regards
Michael
Received on Thursday, 6 November 2003 12:13:24 UTC