- From: Ralph R. Swick <swick@w3.org>
- Date: Mon, 06 Sep 2004 20:45:12 -0400
- To: www-html-editor@w3.org
- Cc: public-rdf-in-xhtml-tf@w3.org
Comments on http://www.w3.org/TR/2004/WD-xhtml2-20040722/ I read only sections 19 Metainformation Attributes Module and 20 Metainformation Module and followed some of the references so if any of my questions are answered in other sections please accept my apologies and consider this a suggestion to add more cross-references. Overall I am excited to see these two sections proposed for XHTML. These two modules have the potential to permit integration of RDF graphs into XHTML documents in a way that should be a relatively small learning curve for XHTML users. I am disturbed, however, by the lack of explicit connection to RDF semantics. The module is eminently suited to expressing RDF semantics, and to use RDF to give precise interpretation of the semantics of the new XHTML vocabulary, but sadly this opportunity has not been taken. It ought to be straightforward to say that the URI that is the value of a property attribute corresponds to the URI that names an RDF property, that the URI that is the value of an about or resource attribute corresponds to the RDF subject and that the metadata value, whether specified by the content attribute or by element content, is the RDF object of a statement. This would then make it possible to define a normative GRDDL [1] transform from this metadata syntax to an RDF graph. I hope this can be rectified in a future Working Draft. [1] http://www.w3.org/TR/grddl/ In Section 19.1, in the second example for the rel attribute, the example shows the media attribute. Does this attribute have any intended metadata semantics? While the other attributes have a relatively obvious representation in an RDF graph (though that representation should be made explicit), it is not clear what representation the media attribute might be intended to have. In Section 19.1 in the description of the restype attribute, the words "The user agent must combine ..." are used. I recommend that this be changed to 'should combine', as there is no apparent need to require user agents to understand the semantics of this particular piece of the metainformation module. If a user agent fails to intersect the restype value with its own acceptable media types, the server will still behave in an appropriate manner. Section 19.3 refers to the property xhtml2:reference. Is this property further defined somewhere? I note that this example shows that the value of this property (e.g. the rdfs:Range) can be either a string or a resource. Section 20.1 uses the URI value http://example.com/people/MarkBirbeck/654 to refer to a person. For pedagogical reasons [2,3], it would be good to use URIs not containing a '#' only to refer to documents and to use a URI such as http://example.com/people/MarkBirbeck#654 to refer to the non-document. The same change should be made to the examples in section 20.3.2. [2] http://www.w3.org/DesignIssues/HTTP-URI.html [3] http://www.w3.org/TR/webarch/#fragid Section 20.2.1.1 shows two examples of property names that do not have namespace qualifiers. To be consistent with other similar uses, the reader should interpret these properties as being in the xhtml2 namespace. I doubt that is what was intended. I recommend that a well-known namespace such as Dublin Core be used in this example. A similar problem occurs in section 20.2.1.3. In that section, however, the intent is to show a not-recommended usage. So I suggest that the example include an XML comment saying that this example should not be followed. Section 20.2.2 refers to the href attribute. What correspondence is there between the href attribute used in this case and the resource attribute used elsewhere? Sectio 20.2.3 refers to 'lists' of property names. I suggest that the term 'vocabularies' conveys your intent and would be understood both by readers familiar with HTML and readers familiar with RDF. Sections 20.3.2 and 20.4 were damaged in your Postscript and PDF versions. The second example in section 20.4 uses a property QName whose namespace has not been shown to be declared (con:). The explanation of this example would be more precise if it where to read as "1. The quote has a source identified by ISBN. (This ISBN number is assigned to the book Crime and Punishment, though this metadata example does not supply that particular information.)" I am confused by section 20.5.2. The text says that the value of the title for the French manual is specified as being in the French language, however the French example is not obviously different from the other 3 examples. I suspect that all four examples are intended to specify the (XML) language of their respective title values. I suggest that the example at the end of section 20.6 is misleading. It says that a string value has identical semantics to the value of a resource attribute. I suggest that it is unwise to specify that any string value that has valid URI syntax should be interpreted as identifying a resource. Regards, -Ralph Swick W3C/MIT
Received on Tuesday, 7 September 2004 00:46:47 UTC