- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Thu, 16 Apr 2009 08:32:39 +0100
- To: RIF WG <public-rif-wg@w3.org>
I meant to abstain on last night's vote on lists (I claim befuddlement due to anti-social hours :-)). This doesn't change the outcome at all but I wanted to formally register my reasons for abstaining. Lists are clearly useful and used in most production rule and LP engines so having them in both PRD and BLD and thus in Core makes sense. However, for RDF rule engines including Jena (and I believe CWM and Euler) there is a small problem. We already have lists but they are RDF lists. An RDF engine shouldn't implement RIF lists by translating them to RDF lists because that would generate additional frame entailments. Technically it would be a small matter to extend such systems to implement RIF lists as well, it violates "implement by translation" but we've violated that already. However, from a user viewpoint this would be a problem - there would now be two different list constructs with different characteristics and we know that anything that confuses users leads to support costs (much more important than implementation costs). Thus there is a risk that Jena would not implement these Lists or would do so via translation to RDF lists and either way would not be a candidate Core implementation. C'est la vie Dave
Received on Thursday, 16 April 2009 07:33:24 UTC