Re: Example problem with QNames and RDF was: Re: QName Problem Isn't One

Hi Jonathan,

Thanks for finally bringing some concrete examples to the table. I'll try
and explain why I don't feel they are a problem.

Jonathan Borden <> wrote:

> <xsl:template match="xsd:decimal">
>   <xsd:decimal>
>       <xsl:value-of select="1 + number(text())"/>
>   </xsd:decimal>
> </xsl:template>

This example is irrelevant because I don't believe it will work under any
XML Schema system.

SOAP uses something like this:
    <foo xsd:type="decimal">1</foo>

Most XML has types defined in the schema.

RDF uses something like this:
    <foo><xsd:decimal rdf:value="1"></foo>

None of these would be properly handled by your example XSLT, and I cannot
think of something that would.

We have never expected different XML formats to be strictly interoperable --
XML allows people the freedom to define their formats in many different (and
incompatible) ways. I don't see why we should hold RDF to a higher standard
than any other XML-based format.

> My response is: if you care about interoperability with XML and XML Schema,
> then you need the QNames to match because QNames are how datatypes are
> refered to in XML, and QNames are how elements and attributes are matched in
> XPath and on and on.

And my response is that I know of now XML-based formats that use such Qnames
for elements or attributes.

> If you don't care about XML compatibility, then why use XML Schema
> datatypes? What to people hope to accomplish by this in the absence of using
> XML and XML related software?

XML Schema has done us the work of defining a recommendation-track set of
datatypes. I see no reason for us to duplicate their effort, but instead we
should try to use it within RDF in the most compatible way we can.

[ Aaron Swartz | | ]

Received on Tuesday, 8 May 2001 19:08:58 UTC