- From: Gary Hallmark <gary.hallmark@oracle.com>
- Date: Mon, 13 Oct 2008 21:59:43 -0700
- To: Christian de Sainte Marie <csma@ilog.fr>
- CC: RIF WG <public-rif-wg@w3.org>
> Rather than focus on the semantics I proposed initially for Assert and > Retract, which I hoped was correctand useful, but which was meant > mostly to start the ball rolling, I propose that we discuss and decide: Why don't we focus on the semantics that I propose, which complements Adrian's changes to use a model theory for conditions? Basically, my proposals involve changing the semantic structure according the action. It seems like we need to nail this down before we talk about more actions. Surely new, assert, and retract are a good starter set of actions. I propose we stick with these and precisely define their semantics for WD2. I find it rather annoying that Adrian and I have made major changes (I think in a good direction) but there are still many rough patches and lots to do to make it ready to publish, yet we aren't discussing these changes and assigning actions to fix issues that are in the spec now. > - first what are the actions we *need* to be able to represent in PRD; > - then, what set of action primitives (atoms) is the most *useful* to > represent all of them; > - last, what is the most appropriate concrete RIF-PRD syntax for these > primitives. > > To start the first step, I started a wiki page [1], where I listed the > elementary actions that correspond to the action primitives in OMG PRR > OCL and in the current draft of PRD. I don't know exactly what you mean by the terms "object", "engine", "scope", "variable", or "members". These terms either aren't in the current PRD spec, or they don't seem to be used in the same way. Why can't "Assign a (new) value to a member of an object (in the scope of the engine)" be translated to ?obj[member->value] :- naf(Exists ?old (?obj[member->?old])) > > I invite you to complete the table by adding a column for you system, > and checking those actions that it is able to represent (e.g., with > some indication of the syntax), and by adding rows for any additional > actions that your system is able to represent. Will this let us make progress on WD2, only 3 weeks away? I think it would be better to formalize the semantics and then ask folks to review it in light of implementability of a translator to/from their system. > > The completed table would provide useful input to the discussion. > > [1] http://www.w3.org/2005/rules/wiki/PRD_Actions > > Cheers, > > Christian > > > >
Received on Tuesday, 14 October 2008 05:00:33 UTC