See also: IRC log
<ChrisW> Agenda: http://lists.w3.org/Archives/Public/public-rif-wg/2006Nov/0052.html
Christian: Next meeting is next week.
Christian: actions 164 done, 165, 171 also done.
<ChrisW> action-165 and action 171 continued
<ChrisW> PROPOSED: Approve 24/10 telecon minutes
<ChrisW> [http://lists.w3.org/Archives/Public/public-rif-wg/2006Nov/att-0002/meeting-2006-10-24.html]
Christian: minutes of Oct. 24 accepted.
<ChrisW> RESOLVE: accept Oct 24 minutes
<ChrisW> RESOLVED: accept Oct 24 minutes
Christian: minutes of Oct. 31 approved.
<ChrisW> PROPOSED: Approve 31/10 telecon minutes
<ChrisW> [http://lists.w3.org/Archives/Public/public-rif-wg/2006Oct/att-0126/rif-minutes-31-Oct-2006.html]
<ChrisW> RESOLVED: accept Oct 31 minutes
Christian: Amendments to the agenda? None.
Christian: move to item F2F meetings. F2F 5 will not be already in Jan.
Sandro: Still no answer on room at MIT for F2F 5.
<ChrisW> ACTION-121 continued (reworded)
<ChrisW> ACTION-122 continued
Deborah: I have no material from Allen on room available for F2F 5.
Christian: Allen told me he booked a room for 3 days.
Deborah: no additional room available.
Sandro: Which dates.
Chris: 26, 27, 28 Feb 2007.
Christian: either 3 days or 2 days.
<ChrisW> Possible dates for next f2f: McLean, VA. Feb 26-28
Christian: let us decide on dates for F2F 5 asap. Are there other proposals?
Chris: Let us give up till 21 Nov for further proposals.
Christian: Decision on place for F2F 5 in 2 weeks from now. Registration forms should be ready next week.
Sandro: Feedback desirable on whether people like dates for F2F.
Christian: This has been already discussed at last F2F.
<ChrisW> ACTION: christian to send email about f2f5 process [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action01]
<rifbot> Created ACTION-183 - Send email about f2f5 process [on Christian de Sainte Marie - due 2006-11-21].
Chris: At last F2F people did not wanted to have F2F in Jan but in Feb, there has been only one proposal for location, as already mentioned.
Christian: move to item liaison.
Christian: any reports from liaison? We could change liaisons because of rare attendance.
<PaulVincent> PRR: submission deferred to next year
Chris: let us defer decision on XQuery liaison.
Christian: Final draft differed to 3 months from now.
Sandro: current state of RIF being able to see what PRR is doing?
Chris: we got permission from OMG to see draft.
<csma> http://www.w3.org/2005/rules/wg/wiki/FrontPage?action=AttachFile&do=get&target=PRR-draft-010606.pdf
Christian: it is on the RIF WG wiki as mentioned above.
Christian: move to item technical design. On the F2F wiki Chris copied resolutions taken at last F2F. We also have a large number of actions.
Christian: action on Hassan to produce an implementation of RIF.
<Hassan> continued
Christian: Hassan's action continued.
<ChrisW> action 142 continued
Christian: Laura rule example for UC 6?
Laura: action continued.
<ChrisW> action 159 continued
<PaulaP> continued
Christian: action on Paula examples for UC 2? continued.
<ChrisW> action 160 continued
<PaulVincent> UC1 continued
Christian: actions on Paul examples for UC 1? continued.
<ChrisW> action 152 continued
<ChrisW> action 153 continued
Christian: action on Allen examples for UC 3: continued.
<ChrisW> action 154 continued
<ChrisW> action 154 completed
<agiurca> he have them
Christian: action 154 done.
<agiurca> we have them. We will put them into wiki
<ChrisW> action 156 completed
Christian: action on Sandro examples on UC 5? continued.
Sandro: does someone to take it over?
<agiurca> http://oxygen.informatik.tu-cottbus.de/rewerse-i1/?q=node/22
<ChrisW> action 156 continued
<ChrisW> action 157 continued
Christian: action 157 on Gary examples for UC 9: continued.
<agiurca> http://oxygen.informatik.tu-cottbus.de/rewerse-i1/?q=node/24
Christian: action on Axel: examples for UC 10? Axel not here.
Chris: Axel's action is done.
<ChrisW> action 158 completed
Christian: actions on Harold to work with Hassan to reconcile proposals: done.
<ChrisW> action 163 completed
Christian: action 166 on Sandro: continued.
<ChrisW> action 166 continued
Sandro: more talk about it needed.
<Hassan> what action is that?
<csma> ack ??P27
Dave: I understand SPARQL group has been speaking about a related issue.
Christian: Dave, have you a reference on this?
Dave: I'll try to find and give it.
Chris: We drop action 166 because it is more an issue than an action.
<ChrisW> action 166 dropped
Christian: action 170 on Alex to list and clarify issues raised by extensibility requirements? Alex not here.
<ChrisW> action 170 continued
Christian: action 170
continued.
Action 182 on Sandro: continued.
<ChrisW> action 182 continued
<ChrisW> action 180 continued
Christian: action 180 on Harold?
Harold: first version ready, action continued.
Christian: Items Forward-Compatibility and asn06.
Christian: Sandro what offer asn06 what EBNF does not offer.
Sandro: role names relates to rule ontology. asn06 is more abstract than EBNF.
<Harold> Sandro, role names can also be given explicitly in the EBNF: they are already in the DTD.
Sandro: asn06 adds meta data.
Francois: meta data to EBNF fine. Stronger statements like asn06 is more abstract than both data type formalisms and EBNF are rather dangerous because they might be difficult to convincingly substantiate.
<PaulaP> +1 to Francois' comment
Christian: people reading the spec will have to understand the formalism.
Harold: EBNF can coexists withg a metalanguage. The meta language should be spearated from the spec. formalism we use.
Sandro: other issues: sequencing, ordering terms, role names, concrete syntax.
Christian: role name is no problem.
Sandro: I think it is a problem. I could not find any proposal with role names.
Christian: EBNF is well understood and well established.
<Harold> Sandro, the DTD has <head> and <body> role names and uses "|" for allowing unorderedness
Christian: Harold could propose an extension to EBNF with role names.
<csma> ask fgm
Frank (fgm): Sandro you dismissed using UML. Why?
Sandro: because we cannot send it in email or on IRC.
Christian: Would be your only problem with UML?
Sandro: I do not know.
Frank: UML has additional features thar are usefull. All kinds of things can be done in UML.
Francois: we should not initiate to much work on side issues. Sandro's points are interesting and surely EBNF can be enriched with role names and other meta data. But designing a fully satisfying language probably would turn out to reqauire more work than it might seem at first. Such an additional issue is not the task of the RIF WG.
Hassan: All theses notations, UML, asn01, asn 06 are often complicated. We need something we can write. Position vs. roles is indeed important. Many other issues are important as well. Defining a API is done at ILOG using CaML what shows that EBNF and the like often do not suffice. I like Sandro's proposal because it can be mapped to LIFE, to EBNF, etc.
<Harold> Sandro, basically <!ELEMENT Implies ((head, body) | (body, head)> can be used, where <!ELEMENT head (%LITFORM;)> and <!ELEMENT body (%CONDIT;)> (cf. http://www.jdrew.org/rif/HornRules.dtd).
Christian: Resistance is againt yet another notation that might make proposal difficult to grasp.
Hassan: If it can be mapped, no problems.
<agiurca> May be is better to keep both EBNF and UML for a while
Christian: If we do not want to waste time, can we decide to start with EBNF with annotations?
Several Attendees:+1.
Gerd: If we start with EBNF, this ok. But we should strive for an EBNF style in the spirit of Sandro's proposal.
Sandro: EBNF is no standard.
Gerd: asn06 is hardly recognizable because it is yet another formalism
Sandro: there are other issues.
<Zakim> ChrisW, you wanted to propose a compromise
Sandro: for adding things EBNF prductions must be modified.
Chris: No need to decide right now. Sandro could continue maintaining asn06, other could gain experience with it. No decision needed at this point.
Christian: This was my proposal.
Hassan: If Sandro can make it more explicit?
Sandro: I would like to understand the various issues expressed by Francois and Gerd.
Christian: Michael, who is not here today, also has provisos.
Gerd: We should not duplicate syntaxes.
<PaulaP> Michael and Axel also sent comments on the proposal
<Harold> BTW, finding a better metasyntax (not 'abstract syntax') is relevant for most W3C specs, not only for RIF.
Christian: Let us investigate issues 'parts of rules', 'how they are related', etc. Let us investigate how these issues relate to extension/extensability before deciding on using EBNF, asn01, asn06 or anything else.
<Harold> Perhaps a topic of collaboration between W3C and OMG?
Christian: Urgent remaks on
asn06? none. Decision we all continue to exchange views on
notations and Sandro continues to clarify asn06. Decision on
formalism to be used will be taken later.
Was there anythinhg in asn06 on the content of abstract
syntax to be discussed right now? No.
Second proposal on extensibility.
Sandro: I think the requirement is that RIF should be designed so as to be forward compatible. Questions are consequences if content (from extensions) is ignored.
Paul: Sandro says in his comments "implementation can be forward compatible". We must be precise on forward compatibility.
Christian: We must required that unknown extensions do not break extensions that do not understand them. This should apply at least for standard extensions.
Sandro: I think we can say more.
<PaulVincent> +1
Francois: let discuss the issue on concrete cases, not abstractly.
Sandro: Presentation are things like names of variables. In case of negation, I mean "if a and not be then c". If you do not understand not, you might get the wrong result.
<Harold> An extension which affects the semantics, but ignoring it wont give you any incorrect results -- just fewer results"
<Harold> This has been studied as 'incompleteness'.
SAndro: all open formats have to proceed this way: ignore what they do not understand and proceed.
<fgm> in the context of closed world, incomplete == wrong
<Hassan> I agree with Sandro - his scheme makes sense to me
Christian: The point is what an application is expected to do when it encounter smething it does not understand.
<Harold> An extension which affects performance, so you'll get the same
<Harold> answers/actions, but the performance will be significantly worse
Christian: The point is about telling applications what to do when they do not understand.
<Harold> This could be called 'pragmatics'.
Frank: There are other things that can be done. The choice of behavior in case one does not understand could be specified at the agent level. I do not like thinking of systems processinbg rules they do not understand.
Sandro: There is a class of users who will proceed if they do not understand, other will stop.
Frank: annotation could be added to rules themselves instead of to the language. There is an ontology of things you can do (if some contructs are not understood).
Christian: action on Frank to propose extensions to Sandro's proposal.
<ChrisW> ACTION: francis to propose extension to Sandro's proposal for extensibility at the level of rules, rulesets [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action02]
<rifbot> Created ACTION-184 - Propose extension to Sandro\'s proposal for extensibility at the level of rules, rulesets [on Francis McCabe - due 2006-11-21].
<Harold> "... an extension which offers syntactic sugar."
Frank: distinction between syntactic sugar, presentation and semantics is harder than you think.
<Harold> This is sometimes called 'conservative extension'.
Christian: Can you give example in email?
Frank: people use conventions for variable names. They often are needed for understanding rules.
<Harold> E.g., if the source language has n-ary lists, but the target language doesn't, but does have complex terms, then the lists can be 'simulated
<Harold> in the the target language via 'cons' complex terms.
Paul: Francois seems to think of another kind of forward compatibility. My concern is updating domains in production rules. This is different issue. Forweard compatibilty is very important in practice. Forward compatibilty is a must in commercial world.
Harold: extensions that affect semantics, fewer results refers to incompleteness. There are 3 aspects: syntax, semantics and pragmatics.
Hassan: I agree with Sandro. There a 3 points. Remember inconsistent predicates in a paper by Pascal at RuleML 2006. In practice, they igonre them and it worked. Why not to provide it as an extensibility machanism if it works? Another thing. If you have a system that does not understand negation but there is a proof system that can do it. All this is important. Ignore features might make sense. Delegation to proof mechanisms are meaningful and important.
Christian: I propose to first get an idea of the components of a rule, e.g. ehat harold will propose tommorrow. SEcond, examples of forward compatibility are needed, expressed in Harold's formalism. Who wolunteers to work ourt such examples for next week?
Hassan: What about Pascal's example at the RuleML 2006?
Christian: action on Hassan to explain Pascal's example?
Hassan: I do not have time. I can send an email to Pascal.
Christian: more examples?
<ChrisW> ACTION: hassan to send email to Pascal about his example from RuleML [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action04]
<rifbot> Created ACTION-185 - Send email to Pascal about his example from RuleML [on Hassan Ait-Kaci - due 2006-11-21].
<Harold> We already have one fwd compatibility example: the current CORE contains only one kind of constants (we were asked to keep it that simple, eg for Prolog compatibility). The 'bipartitioned' extension splits constants into individuals vs. data.
Harold: We should split constants into individuals instead of data.
Christian: action on Harold to do this?
<ChrisW> ACTION: harold to provide example of how his proposal could deal with uniform constants, etc [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action05]
<rifbot> Created ACTION-186 - Provide example of how his proposal could deal with uniform constants, etc [on Harold Boley - due 2006-11-21].
Harold: constants become individual vs. data.
Christian: The action on Harold is on
this very issue.
Any more remark on forward compatibility? No.
Christian: let us have a quick look at other topics.
<ChrisW> action-148 continued
<ChrisW> action-149 continued
<ChrisW> action-172 continued
Christian: action review RIFRAF.
Gary's ACTION: continued. Paul's action:
continued. Sandro's action: continued.
... Laura's ACTION: continued.
<ChrisW> action-173 continued
<ChrisW> action-175 continued
Christian: Allen's ACTION: continued.
<ChrisW> action-176 continued
Christian: Sandro's continued.
<ChrisW> action-177 continued
<ChrisW> action-178 continued
Christian: Axel's action given to Frank? yes.
<ChrisW> action-178 complete
<ChrisW> ACTION: frank to revise and ontologive Section 6 [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action06]
<rifbot> Sorry, couldn't find user - frank
<PaulaP> continued
<ChrisW> ACTION: francis to revise and ontologive Section 6 [recorded in http://www.w3.org/2006/11/14-rif-minutes.html#action07]
<rifbot> Created ACTION-187 - Revise and ontologive Section 6 [on Francis McCabe - due 2006-11-21].
Christian: action 178 done. Paula's two actions continuied.
<Hassan> contineued
Christian: action on Hassan continued.
<ChrisW> action-174 continued
Christian: anything to add on RIFRAF? No.
Christian: no urgent action on UCR? Yes on UC 4 and 5.
<ChrisW> action 144 continued
<ChrisW> action-38 continued
<ChrisW> action-41 continued
Christian: action on me continued.
Christian: no urgent matters. We adjourn.
Several Attendees:+1