- From: Hausenblas, Michael <michael.hausenblas@joanneum.at>
- Date: Sat, 3 Mar 2007 13:39:16 +0100
- To: "Dan Connolly" <connolly@w3.org>
- Cc: <www-qa@w3.org>, <public-rdf-in-xhtml-tf@w3.org>
Dan, > Look at it this way: if I just build a piece of > software that hard-codes all the answers to the > tests, it will pass all the tests, right? But > it won't actually do anything useful for inputs > other than the ones in the test suite. Thanks for this argument, though IMHO one of us two starts with an incorrect precondition. Let me show you what I mean: The rational behind the Test Suite (TS) - that is btw the name, already ;) - is to show how a conforming RDFa implementation should behave, viz. the TS is a set of necessary, but not necessary and sufficient conditions. Put in more formal words, partially using the terminology from [1]: conformantToRDFaSpec(?x) <- RDFaAwareAgent(?x) and _not_ conformantToRDFaSpec(?x) <-> RDFaAwareAgent(?x) And, even if I am wrong with the above, my feeling tells me that there is some wrong with your argumentation. This is due to that in your explanation - as far as I understand it - conformance testing (in finite time) wouldn't be possible at all ;) Cheers, Michael [1] http://www.w3.org/2006/07/SWD/RDFa/#sec3 <http://www.w3.org/2006/07/SWD/RDFa/#sec3> ---------------------------------------------------------- Michael Hausenblas, MSc. Institute of Information Systems & Information Management JOANNEUM RESEARCH Forschungsgesellschaft mbH Steyrergasse 17, A-8010 Graz, AUSTRIA ---------------------------------------------------------- ________________________________ From: Dan Connolly [mailto:connolly@w3.org] Sent: Fri 2007-03-02 22:52 To: Hausenblas, Michael Cc: www-qa@w3.org; www-qa-wg@w3.org Subject: RE: misuse of "conformance testing"? On Fri, 2007-03-02 at 22:33 +0100, Hausenblas, Michael wrote: > Dan, > > Thank you for this hint. I have to admit that I am > not sure 100% by myself; especially regarding QA issues, > I'm always open to learn, though. > > Now, IMHO our intention was to create something like > a conformance test, viz. defining test assertions on > which one can test if her implementation _does_ conform > to RDFa. That's a *very* high bar. I wonder if you really mean to meet it. I'm sure you mean to make a test suite that can help to find lots of ways that an implementation does *not* conform. But do you really mean for the test suite to serve as a *complete* specification of the language, such that the *only* way to pass all the tests is to build a conforming implementation? Look at it this way: if I just build a piece of software that hard-codes all the answers to the tests, it will pass all the tests, right? But it won't actually do anything useful for inputs other than the ones in the test suite. > This is the focus of the Test Suite rather than > i14y, performance, etc. > > In case you've got some better terminology, let us know ;) Just call it a test suite. And say that it's purposes are (a) to clarify issues that come up during the specification development and (b) to assist with test-driven development of implementations. -- Dan Connolly, W3C http://www.w3.org/People/Connolly/ D3C2 887B 0F92 6005 C541 0875 0F91 96DE 6E52 C29E
Received on Saturday, 3 March 2007 12:39:24 UTC