W3C home > Mailing lists > Public > www-rdf-interest@w3.org > March 2002

RE: parseType="literal"

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Tue, 12 Mar 2002 10:50:22 -0000
To: <tarod@softhome.net>, "Danny Ayers" <danny666@virgilio.it>
Cc: <www-rdf-interest@w3.org>
Message-ID: <JAEBJCLMIFLKLOJGMELDOEEPCDAA.jjc@hplb.hpl.hp.com>
>   I still don't know why there are so many problems...

It's XML's fault :)

XML does not fully support embedding of one XML document within another.

>   First of all, rdf doesn't allow name of resources like "q:name", or it
> accepts it but it doesn't translate the q to the namespace defined,

Correct, we are only talking about qnames in attributes valeus in xml
literals in RDF.
(Fairly specialised!).

>                                                               but if
> you add xml in your literals, you need to know what namespaces are using,
> etc...


>           why don't we leave the literal as it is,

Running above an XML parser, this is not possible. Minimally things not in
infoset are lost.

Running above an XPath Nodeset based XML processor (e.g. XSLT) more than
that is lost, including some namespace declarations.

>                                                when user loads this
> literal he will try to read it and if he doesn't have this namespace
> defined it will fault.

One possible choice is that namespaces get transmitted independently of the
xml literal. Then if the namespaces are not defined it is the document
user's fault.

Another possible choice, is that the namespaces must be declared within the
xml literal. Then again if they are not, it is the document author's fault.
That choice is attractive, except that it prohibits an XSLT based
preprocessing of RDF/XML.

>   By the way, why RDF doesn't interpret the namespaces in the attributes
> values?? will it do in the future? I don't understand why it
> doesn't do it.

RDF2 is likely to in my view.
No one argues against it.

RDF M&S (and hence the new specs) do not allow it because XML did not allow
it at the time.

>   My proposal is to leave the literals as strings, when the user ask for
> this string, rdf should return this string, the user must interpret is as
> xml and fail if there is any error in the xml.

This proposal is in opposition to the desire to permit processing of RDF/XML
by standard XML tools (such as a SAX2 parser). The strings include things
not in infoset and hence not accessible from XML apis.

>   Why?
>   Imagine I want to define my views in rdf, so I will have something like
>   <rdf:rdf>
>      <own:page rdf:parseType="Resource">
>         <own:header rdf:parseType="Literal">
>              //xml tags to create our page (xul)
>         <own:header>
>         <own:content rdf:parseType="Literal">
>              // my content in xul
>         <own:content>
>         <own:foot rdf:parseType="Literal">
>              //xml closing the tags opened in header (xul)
>         <own:foot>
>      </own:page>
>   </rdf:rdf>
>   I don't want to process the info in the resources, I want to create a
> full document and send it to mozilla.
>   Is it possible? doesn't anybody else need this functionallity?
>   Thanks,
>          Marc
Received on Tuesday, 12 March 2002 05:50:39 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:39 UTC