Re: [PRD] PRD TF telecon Tuesday 17 February

Paul Vincent wrote:
> 
> Comments on discussion:
> - Gary's example [2 below]: neither of these rules will fire as there is
> no such thing as (or intent for) a (PRD) RIF rule engine! Therefore the
> complaint is moot until we include expected translations to PRD targets
> (which I think Gary has assumed). Ergo this is not so much a problem
> with PRD syntax and semantics, but in the expected translators to such
> PRD syntax. Having said that, it might be that this comment is trying to
> say that PRD cannot well handle situation X... which would be a problem
> for RIF.

Indeed, I think that Gary's point was that the current specification of the default conflict resolution strategy requires that implementations translate the rules in such a way that the first one will fire only once and the second twice, with the same, given, initial state of the working memory.

Since most rule engines will handle both rules the same way, I believe, like Gary, that this is a bug in the specification.

I think that this bug can be corrected by modifying slightly the definition of refraction, so that action variables be taken into account iff  their initialisation template matches a ground condition element (in the condition or in a binding pattern) that is not inside an existential condition (in the current version of the spec, only rule variables are taken into account for refraction).

> - FIC debrief [1 below] and Carlos' concern [3 below]: obviously the
> closer the RIF PRD model is to target language terminology (and OMG PRR)
> the better, IMHO. However, I'd long thought this would need to be an
> additional dialect layer (eg PRD-OO or somesuch) after PRD 1.0. 

I wonder if we have the same Carlos' concern in mind... The one I had in mind for the discussion on conflict resolution, was that refraction+priority+recency would leave many conflict unresolved, and that the execution by different engines of the same ruleset in the same initial state of the working memory would thus give significantly different results in too many cases.

Is this what you believe should be an additional dialect layer after PRD 1.0?

Cheers,

Christian


> Cheers
> Paul Vincent 
> +1 650 206 2493 / mobile +44 781 493 7229 
> 
> 
>>-----Original Message-----
>>From: public-rif-wg-request@w3.org
> 
> [mailto:public-rif-wg-request@w3.org]
> 
>>On Behalf Of Christian de Sainte Marie
>>Sent: 16 February 2009 19:14
>>To: RIF WG; Serrano-Morales, Carlos A; Berlioz-Matignon, Carole Ann
>>Subject: [PRD] PRD TF telecon Tuesday 17 February
>>
>>Teleconference W3C RIF WG
>>Production rules dialect (PRD) task force
>>17 February 2006
>>
>>North America and Europe are in the Winter time:
>>1800 UTC, 1000 (West US) 1300 (East US) 1800 (London) 1900 (Paris)
>>
>>Duration: *60 min*
>>
>>*Proposed agenda*
>>
>>1. Debrief PRD meeting with ILOG, FIC and Tibco, Feb 11 [1]
>>
>>2. Discussion on the default conflict resolution strategy
>>- Gary's problematic example [2]
>>- Carlos remark that the current definition is not discriminating
>>enough (but-last item in [3]): shall we add another rule before the
>>tie-breaker; if yes, which?
>>
>>3. AOB
>>- Next meeting: February 24
>>
>>[1]
> 
> http://lists.w3.org/Archives/Public/public-rif-wg/2009Feb/0079.html
> 
>>[2]
> 
> http://lists.w3.org/Archives/Public/public-rif-wg/2009Feb/0054.html
> 
>>[3] http://www.w3.org/2005/rules/wiki/PRD_F2F_Wednesday_11_February
>>
>>*Telecon details*
>>Zakim bridges: +1.617.761.6200 (US), +33 4 89 06 34 99 (F) or
>>+44.117.370.6152
>>(GB)
>>Conference code: 743773 ("RIFPRD")
>>IRC Chat: irc:irc.w3.org (port 6665), #rif-prd
>>Web-based IRC (member-only): [http://www.w3.org/2001/01/cgi-irc]
>>
>>Please note that RIF-PRD telecons are for attendance only by RIF
>>Working
>>Group Participants and guests invited by the chairs.
>>
>>
>>
>>
> 
> 
> 
>

Received on Tuesday, 17 February 2009 14:18:16 UTC