W3C home > Mailing lists > Public > public-grddl-comments@w3.org > January to March 2007

More on GRDDL tests (apology)

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Mon, 22 Jan 2007 10:31:04 +0000
Message-ID: <45B49268.5060803@hpl.hp.com>
To: public-grddl-comments@w3.org
CC: "McBride, Brian" <brian.mcbride@hp.com>, jena-devel <jena-devel@lists.sourceforge.net>

Sorry about the over-zealous virus checker ....
The attachments are now in a zip.

Not a formal comment.

    A. Description of my code.
    B. Publish new WD?
    C. Failing tests - I think issues with test data and not with my code.
    D. Code location

A. Description of my code.

My code now implements what I think is what is currently intended.

1. apply GRDDL transforms declared directly in input
2. if input has a root element in a namespace NS
       if NS is RDF/XML namespace or of mimetype application/rdf+xml,
         also read document untransformed as a GRDDL transform
       else for *NS namespaceTransformation URL* triple in any
            GRDDL transform of NS apply transform URL
3. if input is xhtml (through mimetype or schema or guesswork) then
        for each profile P
           for each
              *P profileTransformation URL*
             triple in GRDDL transform of P
                 apply transform URL
4. if document is of mimetype application/rdf+xml also
     read document untransformed (unless covered under 2)

The code also accept text/rdf+n3 as a transform output media-type.
The intermediate document is streamed from saxon8 into the appropriate
Jena reader (as SAX events for RDF/XML, and as Unicode String for N3),
in principle this should mean that some very large documents can be
handled. The code supports XSLT 2.0.

B. Publish new WD?

I note that the currently published WD adds transform results of
NS and P to the GRDDL result of the original document. If this is no
longer the intent, this seems like a big enough change to merit
consideration of a new WD.

C. Failing Tests

I fail the following tests:
a)   sq1ns.xml => sq1-output.rdf
b)   card.html => card-output.rdf
c)   sq1.xml => sq1-output.rdf
d)   sq2.xml => sq2-output.rdf

e)   loop.xml => loop-output.rdf

I note that sq1ns.xml is not served with mimetype application/rdf+xml
despite being RDF/XML.

card-output.rdf seems to have  been generated with an older version of
the transform

I think this is a follow on from the issues to do with sq1ns.xml

sq2-output.rdf has a file URL in it!

e) Actually I pass this, but got a 404 on:

guessing, I think you mean


I've hard coded that guess into my code, and then get a different result!

I've attached my five different result files, in a zip.

D. Code Location

To do:
    - implement Jena RDFReader interface
    - make quiet
    - options, warnings and errors
    - caching (currently a naive infinite cache to avoid loops!)
    - more tests

    - maybe, content negotiation etc.


Received on Monday, 22 January 2007 10:31:40 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:55:02 UTC