comments on GRDDL tests

Hi Dan, et al.

I'm managing to pass some of the tests now.

Enough to have a few minor issues, and one slightly bigger one (at end):

1) I get a single triple incorrect on testlist1.html

Starting http://www.w3.org/2001/sw/grddl-wg/td/testlist1.html
http://www.w3.org/2001/sw/grddl-wg/td/aboutTests:
GRDDL
=====
<http://www.w3.org/2001/sw/grddl-wg/td/testlist1.html#base-param> 
<http://purl.org/dc/elements/1.1/title> "#base-param:" .

RDF/XML
=======
<http://www.w3.org/2001/sw/grddl-wg/td/testlist1.html#base-param> 
<http://purl.org/dc/elements/1.1/title> "#base-param: the base URI for 
the result document is the URI of the source document." .

2) testlist2.html and testlist3.html don't declare a profile, and so 
don't transform

3) atom-grddl.xml

I don't attempt to pass this yet, and would find it a lot easier if the 
xsl was changed from:
http://www.w3.org/2001/sw/grddl-wg/td/atom2turtle_xslt-1.0.xsl

   <xsl:output method="text"/>

to

   <xsl:output method="text" media-type="text/rdf+n3" />

It wouldn't harm if some of the other tests explicitly had:

   <xsl:output  media-type="application/rdf+xml" />

4) The sq1ns.xml test is inconsistent with the pending projects.rdf => 
projects.rdf test. My code currently assumes a file
with root element in the RDF namespace can be read as RDF/XML. I will 
extend this to be that a document served as application/rdf+xml can be 
read as RDF/XML. If the WG is undecided about this issue, I am happpy to 
wait, and I have no particular opinion one way or another about the root 
element; (if the server serves application/rdf+xml then I will feel free 
to apply an RDF/XML parser).

I am not yet attempting the tests that require processing of a schema or 
profile document ... tomorrow or friday maybe???

5) Bigger issue
I initially got confused, because for some of the tests the input and 
output documents contain different RDF! (Not just the testlist1 
document). The problem is that the files are identical (i.e. the output 
files are exactly what an XSLT transform would produce).
But, the original document is read with one base, e.g.
http://www.w3.org/2001/sw/grddl-wg/td/testlist1.html
and the output document is read with a different one, e.g.
http://www.w3.org/2001/sw/grddl-wg/td/testlist1.rdf
This gives different results.
For that particular case, I passed the test (modulo point 1), by using 
content negotiation to get the two versions of the file from
http://www.w3.org/2001/sw/grddl-wg/td/testlist1

For some of the tests this cannot be done, since they do not have a 
common URL with content negotiation. I am not sure what change I am 
suggesting. In my code, I end up reading the output file as if it were 
from the input file, before comparing the two RDF graphs.

Jeremy

Received on Wednesday, 17 January 2007 15:48:06 UTC