Re: Problematic test cases in the RDFa test suite

Christoph LANGE wrote:
> On Wednesday 17 December 2008 22:49:28 Christoph LANGE wrote:
>> Isn't there a way to rewrite the queries for the test cases 0042, 0086, and
>> 0107?  In an automated setting these tests will, of course, fail on a
>> correct implementation, as a parser should not actually generate any
>> triples. Wouldn't it make sense to query for "?a ?b ?c" instead and check
>> whether none of them is bound?
> 
> Sorry, I realized that there is information about the expected result of a
> test in the manifest file.  That is, any implementation of a test suite should
> actually use the manifest file.

Christoph,

Just to clarify a bit further, we were very careful to check for very
specific triples since a conforming parser MAY generate other triples
when run on those test cases. For example, a processor may generate a
"dcterms:title" for the test case document using the TITLE element to
set the object:

<> dcterms:title "Test Case 0042" .

The SPARQL ensures that the one buggy spec-generated triple that might
be generated due to a bug in the parser is not there.

As for TC0114, are you sure that you don't want any changes to the TC,
even if it is a documentation change? Does your implementation generate
different triples? If so, what are they? As Michael Kay mentioned, RFC
3986 obsoletes RFC 2396, so we tend to favor 3986 for the test suite.

-- manu

-- 
Manu Sporny
President/CEO - Digital Bazaar, Inc.
blog: Bitmunk 3.1 Website Launch
http://blog.digitalbazaar.com/2009/01/16/bitmunk-3-1-website-launch

Received on Thursday, 22 January 2009 15:52:39 UTC