- From: Dan Connolly <connolly@w3.org>
- Date: Fri, 16 Apr 2010 10:01:59 -0500
- To: Chris Welty <cawelty@gmail.com>
- Cc: public-rif-comments@w3.org
On Fri, 2010-04-16 at 10:03 -0400, Chris Welty wrote: > > Dan Connolly wrote: > > On Mon, 2010-03-01 at 12:01 -0500, Chris Welty wrote: > >> Dan, > >> > >> We struggled with this point based on RIF's status as an interchange format, not a rule language per se. Thus the conformance refers to the ability to translate in a way that does not change the semantics, which includes entailments. > >> > >> It is not, as you say, directly observable in a positive way, however it is negatively observable > >> through sets of tests, ie you can test if it did not happen. > > > > I don't see how you could test that it did not happen either. > > > > Do any of the existing RIF test cases show how it can be done? > > Where I'm assuming the "it" here refers to the property of being conformant. Or rather: not conformant. (you can only falsify a hypothesis by experiment, not prove it). > In general, a RIF processor takes statements in RIF syntax and translates them into some > existing rule language. This is described in each of the documents. If you translate a > positive test case and its conclusion into the target rule language, the conclusion > should hold. If you translate a negative test case into the rule language then > the conclusion should not hold. OK, so any of the semantics tests may show that a piece of software fails to be a conforming RIF BLD consumer. I guess that makes sense. Thanks. > > -Chris > > > > >> -Chris > >> > >> > >> Dan Connolly wrote: > >>> I see: > >>> > >>> "A RIF processor is a conformant BLDΤ,Ε consumer iff it implements a > >>> semantics-preserving mapping, μ, from the set of all BLDΤ,Ε formulas to > >>> the language L of the processor (μ does not need to be an "onto" > >>> mapping)." > >>> -- http://www.w3.org/TR/rif-bld/#Conformance_Clauses > >>> > >>> I don't see how this property is observable/testable; i.e. why > >>> this product class is defined at all. > >>> > >>> A conformant RIF-BLD consumer isn't required to compute entailment? > >>> > >>> This much is observable: "A conformant RIF-BLD consumer must reject all > >>> inputs that do not match the syntax of BLD." But that's just syntax > >>> checking. > >>> > >>> editorial: why "conformant" rather than "conforming"? > >>> > >>> > > > > > -- Dan Connolly, W3C http://www.w3.org/People/Connolly/ gpg D3C2 887B 0F92 6005 C541 0875 0F91 96DE 6E52 C29E
Received on Saturday, 17 April 2010 20:53:24 UTC