causality and EARL

Hi, 

  I was wondering what people on this list thought about the following. The WebID XG is working on  specifying an EARL vocabulary to allow WebID authenticators ( latest spec: http://webid.info/spec ) to explain why an authentication succeeded or failed. There are essentially two reasons this can happen:
  1) the X509 Certificate is (not) a good one
  2) the WebID claims in the Certificate can (not) be verified

It is b) that is the most important here. What we have currently defined are 
at least the following tests for b)

  1- the WebID Profile document could be fetched
  2- the WebID profile could be parsed
  3- the SPARQL ASK query on the graph of relations in the Profile succeeded


Now if 3 succeeds then necessarily 1 and 2 succeeded. And if 2 succeeds then necessarily 1 succeeded. So the earl tests for 2 and 3 will always only be worth writing about if they fail. Really 3 is constituted of 1 and 2. Or again 1 and 2 are causes of 3. 

We are also trying to keep the implementation of the test page simple, so that we can explain it easily and get people to put up
these pages spontaneously. What we were wondering is if the following may not be a better way to write things, rather than have 3 tests.


[] a earl:Assertion;
    earl:subject _:certWebIDClaim_1;
    earl:test wit:webidClaim ;
    earl:result [ a earl:TestResult;
        dct:description "Could not parse document";
        earl:cause [ a earl:Error, wit:DocumentParsingError;
                     ... //nothing or something here, perhaps a pointer to something else... 
                    ];
        earl:outcome earl:failed ] .

Is this something that people have already thought of?

Henry


Social Web Architect
http://bblfish.net/

Received on Friday, 2 December 2011 09:38:37 UTC