a few comments

Hi

Brian is encouraging me to think about implementing GRDDL within Jena.

I started by reading the three drafts, and also looked at the editor's 
draft of the main document.

http://www.w3.org/TR/2006/WD-grddl-20061024/
http://www.w3.org/TR/2006/WD-grddl-primer-20061002/
http://www.w3.org/TR/2006/WD-grddl-scenarios-20061002/

http://www.w3.org/2004/01/rdxh/spec

I have not done a detailed review; I could do one now if that would be 
helpful to the group, but I suspect it would help more if I held off 
until either the next publication, or until Last Call.

However I had three immediate comments.

Editorial (praise):
==================

I found the following line in the introductory example highly enlightening:

<foaf:isPrimaryTopicOf 
rdf:resource="http://en.wikipedia.org/wiki/Stephen_King" />

A, somewhat philosophical, issue, that has caused problems to semantic 
web recommendations in the past, is how to relate the formal semantics 
with the real world. I felt that the pointing to the Web that this line 
involves seems to illustrate some of the ideas of

"Clients reading the document can follow their nose " (Primer abstract)

It also seems to illustrate aspects of "Faithful Renditions", which 
perhaps could be made clearer. Typically a faithful rendition will 
involve reference to Web resources which may change under the authors 
feet, and the "faithful rendition" is intended as a best effort, such as 
we typically make when trying to communicate, and isn't going to be 
entirely bullet proof.

Feedback
========
Between "Using GRDDL with an RDF Namespace document" and "Using GRDDL 
with an XML Schema namespace document" there is the comment:

[[
The Working Group is likely to add a section to the GRDDL primer much 
like this subsection. Since this subsection has no novel normative 
material, we're interested in feedback on whether it should remain part 
of this specification once it is covered by the primer.
]]

My own view is that if an example is appropriate for the primer then it 
should be there and not in the specification; however, more complex 
cases, may be usefully illustrated for the developer within (an 
informative) subsection of the specification.

In particular, in the XML Schema case, I understand that multiple GRDDL 
transformations have to be used. This seems like an example of 
appropriate complexity for the spec, with perhaps a simplification in 
the primer that glosses over some of the mechanics.

Related editorial comments
==========================

If I have understood the text correctly then
http://www.w3.org/2004/01/rdxh/figGleanPO.svg
would be clearer if the arrow lablled namespaceTransformation was 
directed from the lower "result RDF" box rather than from the "po-ex" 
box. i.e. the client first applies the embeddedRDF.xsl transform, can 
then read the result to find the RDF indicating the grokPO.xsl transform.

Also "grok" is not a commonly understood word (although I like and use 
it frequently!). Suggest changing "grokPO.xsl" to "understandPO.xsl"


Substantive XSLT 2.0
====================

I was very surprised to see that this spec referred to XSLT 1.0 and not 
XSLT 2.0.
By the time GRDDL goes to rec, XSLT 2.0 is likely to be the W3C 
recommendation, and I don't see why the GRDDL WG should recommend a 
solution that the consortium has significantly improved. In particular, 
browser support seems to be a) irrelevant for many uses (are browsers 
going to be the main GRDDL engines, I think not) b) historically dated, 
browser support for XSLT 2.0 is likely in the future.

Also the minutes of the 30 Aug meeting do not seem to reveal why MUST in 
the e-mail message became SHOULD. While it is clear that some software 
may use a GRDDL mechanism for say processing javascript, and not support 
XSLT, I think it is clearer to say that these are not GRDDL clients.

i.e.

[[
A GRDDL client MUST support XSLT 2.0 and XSLT 1.0

Authors of GRDDL documents are advised that restricting themselves to 
XSLT 1.0 may achieve higher interoperability with clients that partially 
implement the GRDDL specification.
]]

The WG appears to want GRDDL to be an open-ended mechanism for 
experimenting with different transformation languages. I have nothing 
against that, but the recommendation should concentrate on a realistic 
interoperable solution; and the WG should ensure that a client 
implementing the specification can appropriately flag any more 
experimental use of GRDDL that it does not support.


Jeremy

Received on Monday, 4 December 2006 11:58:15 UTC