View source

From RIF

You do not have permission to do that, for the following reasons:

• The action you have requested is limited to users in one of the groups Sysops, WG.
• You must confirm your e-mail address before editing pages. Please set and validate your e-mail address through your user preferences.

You can view and copy the source of this page:

{{chatlog|rrsagent=http://www.w3.org/2008/09/27-rif-irc|chatlog={{fullurl:{{PAGENAME}}}}}} [[F2F11]] Day 2. <pre><nowiki> 00:00:00 <sandro> PRESENT: csma, chrisw, gary, dave, changhai, Adrian, Harold, paul, blaz, michael, stella, mikedean, jos, sandro, axel, said 00:00:00 <sandro> REMOTE: Bob 13:10:44 <DaveReynolds> DaveReynolds has joined #rif 13:12:14 <sandro> sandro has joined #rif 13:13:46 <sandro> RRSAgent, pointer? 13:13:46 <RRSAgent> See http://www.w3.org/2008/09/27-rif-irc#T13-13-46 13:15:52 <GaryHallmark> GaryHallmark has joined #rif 13:16:19 <ChrisW> ChrisW has joined #rif 13:17:38 <mdean> mdean has joined #rif 13:17:44 <Blaz> Blaz has joined #rif 13:17:53 <StellaMitchell> StellaMitchell has joined #rif 13:19:36 <sandro> testing.... 13:19:46 <josb> josb has joined #rif 13:19:50 <sandro> scribe: DaveReynolds 13:20:01 <DaveReynolds> ScribeNick: DaveReynolds 13:20:26 <AxelPolleres> AxelPolleres has joined #rif 13:20:43 <AdrianP> AdrianP has joined #rif 13:21:19 <DaveReynolds> Discussing resolutions proposed from day 1 13:21:34 <sandro> topic: profiles (issue-29) 13:21:36 <ChrisW> PROPOSED: RIF will not specify a mechanism for profiles in RIF dialects 13:21:41 <ChrisW> (closing issue-29) 13:21:43 <Harold> Harold has joined #rif 13:21:53 <ChrisW> PROPOSED: RIF will not specify a mechanism for profiles in RIF dialects (closing ISSUE-29) 13:21:56 <josb> +1 13:22:25 <DaveReynolds> Sandro: clarify that "profiles" means sub-dialects that are not themselves dialects 13:22:59 <sandro> PROPOSED: RIF will not specify a mechanism for profiles (sublanguages of dialects, which are not themselves a dialect), closing issue-29 13:23:16 <DaveReynolds> csma: a mechanism for profiles would be a mechanism for specifying a sub-dialect by restriction of an existing dialect only 13:23:56 <DaveReynolds> Harold: might something like that in CORE 13:24:04 <sandro> csma: eg "BLD-minus-frames" will not be something that folks can define on their own, without us. 13:24:20 <DaveReynolds> csma: we can define things by reference, the point is to not have a generic mechanism 13:24:23 <csma> csma has joined #rif 13:24:31 <DaveReynolds> s/mechanism/mechanism for this/ 13:24:53 <sandro> PROPOSED: RIF will not specify a mechanism for profiles (sub-languages of dialects, which are not themselves a dialect), (addressing part of issue-29) 13:25:23 <sandro> Harold: How is this different from a specialization using FLD? 13:25:34 <DaveReynolds> Harold: difference is that the output is a dialect or not a dialect, the profile lacks that status of a formal dialect 13:26:28 <DaveReynolds> No objections 13:26:35 <sandro> RESOLVED: RIF will not specify a mechanism for profiles (sub-languages of dialects, which are not themselves a dialect), (addressing part of issue-29) 13:27:08 <BobMoore> BobMoore has joined #rif 13:27:10 <DaveReynolds> Next part of issue-29: do we allow dialects which extend only a subset of CORE 13:27:15 <ChrisW> rrsagent, pointer? 13:27:15 <RRSAgent> See http://www.w3.org/2008/09/27-rif-irc#T13-27-15 13:27:18 <sandro> csma: Is CORE the root of all extensions, all dialects? 13:28:18 <DaveReynolds> Sandro: argument against this is that might find later there are things in CORE we find too heavy 13:29:05 <DaveReynolds> ChrisW: still allows the notion of a CORE, but it might have to change, CORE should remain the LCD 13:29:36 <DaveReynolds> csma: yesterday we were close to agreeing to allow extensions of subsets of CORE 13:29:42 <DaveReynolds> Sandro: <surprise> 13:30:11 <PaulVincent> PaulVincent has joined #rif 13:30:15 <DaveReynolds> Gary: can't stop vendors doing some vendor PRD profile that might not include all of CORE 13:30:46 <DaveReynolds> Sandro: but that would not be RIF 13:32:34 <DaveReynolds> ChrisW: yesterday's phrasing was to not constrain ourselves as to whether or not there would ever be a RIF dialect with is not a superset of all of CORE 13:33:01 <DaveReynolds> Sandro: but want to discourage that 13:33:25 <BobMoore> Is the phone bridge active? What is the conference number for today? 13:34:10 <DaveReynolds> Harold: in the latter case the subset of CORE is random, doesn't make sense on its own. If the subset of CORE is meaningful that might be more acceptable. 13:34:22 <sandro> Hold on BobMoore 13:34:29 <DaveReynolds> Harold: must be a community effort, not just vendor 13:34:44 <sandro> zakim, room for 10 for 600 minutes 13:34:44 <Zakim> I don't understand 'room for 10 for 600 minutes', sandro 13:34:46 <sandro> zakim, room for 10 for 600 minutes? 13:34:47 <Zakim> ok, sandro; conference Team_(rif)13:34Z scheduled with code 26631 (CONF1) for 600 minutes until 2334Z 13:35:14 <sandro> dialing in the room now, Bob 13:35:31 <Zakim> Team_(rif)13:34Z has now started 13:35:39 <Zakim> + +7.778.41.aaaa 13:36:16 <sandro> Zakim, +aaaa is BobMoore 13:36:16 <Zakim> sorry, sandro, I do not recognize a party named '+aaaa' 13:36:17 <Zakim> +??P1 13:36:29 <sandro> Zakim, +7.778.41.aaaa is BobMoore 13:36:29 <Zakim> +BobMoore; got it 13:36:36 <sandro> zakim, ?P1 is Meeting_Room 13:36:36 <Zakim> sorry, sandro, I do not recognize a party named '?P1' 13:36:40 <sandro> zakim, ??P1 is Meeting_Room 13:36:40 <Zakim> +Meeting_Room; got it 13:38:08 <DaveReynolds> csma: we could allow ourselves to specify a sub-CORE in future for such purposes 13:38:16 <DaveReynolds> Sandro: let's deal with that we 13:38:22 <DaveReynolds> s/we/when the problem arises/ 13:38:50 <DaveReynolds> Paul: in that case the problem is the definition of CORE, not nec. the dialect that wants the subset 13:39:14 <sandro> PaulVincent, Sandro, Chrisw: If we find we need "D3" (a dialect extending a subset of Core) then we're in a bad situation, which we'll deal with at the time. 13:39:27 <DaveReynolds> Chrisw: should close issue-29 and open a new issue "is CORE required by every dialect" 13:39:41 <DaveReynolds> Sandro: don't need an issue there, if it arises then it becomes an issue then. 13:40:27 <csma> PROPOSED: we do not decide that all future dialect MUST extend Core (closing ISSUE-69) 13:40:29 <DaveReynolds> Chrisw: issues reflect uncertainty or disagreement, the fact that we considered this and didn't conclude should be carried forward 13:41:11 <josb> josb has joined #rif 13:41:13 <csma> PROPOSED: we do not decide that all future dialects MUST extend Core (closing ISSUE-29) 13:42:03 <DaveReynolds> Chrisw: this would be an issue we postponed in order to record for future working groups 13:42:04 <csma> PROPOSED: we do not decide that all future dialects MUST extend Core (postponing ISSUE-29) 13:42:06 <GaryHallmark> future dialects SHOULD extend Core, however the WG recognizes that there is a trade-off between interoperability and ease of translator implementation. Therefore it is not required. 13:42:25 <sandro> ... and mostly, we'll have to deal with it at the time. 13:43:35 <sandro> PROPOSED: Future dialects SHOULD extend Core. (Closing ISSUE-29) 13:43:49 <sandro> PROPOSED: Future RIF dialects SHOULD extend Core. (Closing ISSUE-29) 13:45:24 <sandro> RESOLVED: Future RIF dialects SHOULD extend Core. (Closing ISSUE-29) 13:45:38 <sandro> RRSAgent, pointer? 13:45:38 <RRSAgent> See http://www.w3.org/2008/09/27-rif-irc#T13-45-38 13:45:56 <DaveReynolds> Next proposed resolution was to remove "at risk" on external frames on BLD 13:46:55 <sandro> DaveReynolds: MK mentioned the two issues: (1) where do you send the query, and (2) ...?... 13:47:26 <sandro> DaveReynolds: That's not a motivation -- if you want that, then add a language feature to do that. 13:47:55 <AxelPolleres> q+ 13:48:23 <DaveReynolds> Gary: possible relates to the issue of PRD systems being unable to put membership in the head, because membership is "externally" defined 13:48:58 <DaveReynolds> Gary: dismayed by the asymmetry - external member/subclass more interesting for him than external frames 13:49:35 <DaveReynolds> Gary: that would then explain why it can't go in the head 13:50:05 <DaveReynolds> csma: right, in PRD case they are always external 13:50:55 <DaveReynolds> Sandro: how would that look? 13:51:10 <DaveReynolds> Gary/csma: External( ?x # IRIclasss) 13:51:18 <DaveReynolds> Adrian: better to have typed variables 13:51:28 <DaveReynolds> csma: this is a redesign 13:51:40 <DaveReynolds> Sandro: could consider it an extension 13:51:45 <DaveReynolds> MK: easy to add 13:52:11 <DaveReynolds> MK: just an oversight it wasn't included 13:52:22 <DaveReynolds> Harold: this is in FLD 13:52:34 <Harold> BLD : 'External' '(' Atom | Frame ')' 13:53:03 <DaveReynolds> MK: referenced overnight email 13:53:06 <Harold> FLD : 'External' '(' ATOMIC ')') 13:53:06 <Harold> ATOMIC ::= Const | Atom | Equal | Member | Subclass | Frame 13:53:33 <DaveReynolds> MK: can't know where to address unless you have a convention, irrespective of frames/predicates etc 13:53:46 <DaveReynolds> MK: in this case the convention could be class 13:54:10 <DaveReynolds> Sandro: no, the class doesn't know about the instances 13:54:55 <AxelPolleres> We can do ALL external calls with predicates, e.g. External( a [ p-> c] ) could be formulated as: External( call( "?a [ ?p-> ?c]" a b c ) or an external call to a SPARQL or SQL endpoint could be emulated with External( EvalSPARQL( ?X ?Y " SELECT ?X FROM ... WHERE { ... ?Y ... } " ), etc. 13:55:23 <DaveReynolds> csma: MISMO example: ?x # Mismo:elt 13:55:45 <AxelPolleres> ... so theoretically ther is neither an advantage nor harm in allowing/disallowing whatever terms in External. 13:55:47 <DaveReynolds> csma: the fact that this is external is only given by the fact that Mismo schema exists 13:57:00 <AdrianP> we are talking about an external type system 13:58:05 <DaveReynolds> Dave: talking about three different cases - access to an external data source (c.f. email trail), a builtin with a fixed interpretation and an object where you change the slot values but not the type 13:58:12 <AxelPolleres> zakim, who is on the queue? 13:58:12 <Zakim> I see AxelPolleres on the speaker queue 13:58:30 <DaveReynolds> MK: builtins are external because they are not defined by your ruleset 13:59:00 <DaveReynolds> Changhai: what does that mean in the case of something like Mismo? 13:59:26 <DaveReynolds> Sandro: if it is not external, are you somehow assuming there is no information available elsewhere? 13:59:53 <DaveReynolds> MK: if it isn't external it will try to match against clauses in the ruleset 14:00:45 <DaveReynolds> csma: if you can assert it then it can't be external 14:01:43 <MichaelKifer> MichaelKifer has joined #rif 14:01:50 <DaveReynolds> Axel: we clarified yesterday we could do all external calls with predicates using some query string. 14:02:03 <DaveReynolds> Axel: two ways to go, either only predicates or allow any TERMs 14:03:43 <PaulVincent> Axel: its not about "external calls" (I think), its about using a Domain Specific Language with predicate constraints defined by the schema, and this being used to build the rules, with runtime-only access to facts (when you "run the rules with dataset X")... 14:03:56 <DaveReynolds> csma: is x = y, when x y are integer the same as numeric equals or not 14:03:59 <csma> q? 14:04:05 <csma> ack axel 14:04:31 <DaveReynolds> Harold: re Axel's approach, this would be some sort of builtin 14:04:55 <DaveReynolds> Axel: yes, we just have a choice of predicates or all TERMs 14:05:17 <DaveReynolds> MK: in that case the extreme would be to have one builtin called "external" ! 14:06:05 <DaveReynolds> csma: how to distinguish between a membership relation defined externally and the case where you can define or modify it 14:06:12 <AxelPolleres> I propose the following: 14:06:49 <AxelPolleres> Change the Definition of External schemas (sec 2.5 http://www.w3.org/2005/rules/wiki/FLD) as follows: 14:06:54 <DaveReynolds> MK: there is no issue, if you write external, you go to some place based on some convention and the external source will satisfy the goal 14:07:26 <AxelPolleres> to allowing ANY term for &tau; 14:07:31 <DaveReynolds> MK: it doesn't stop you defining something internally as well 14:07:54 <DaveReynolds> MK: ?X:abc :- External(?X:abc) 14:10:27 <sandro> DaveReynolds: CSMA, I think you're confusing two things in your example -- use subclass instead of member, since subclass *is* fixed by MISMO. 14:11:49 <sandro> I'm finding what MichaelKifer said makes sense if I think of "external" as "evaluable" -- but not if I think of external data sources which you could merge with. 14:12:49 <DaveReynolds> [not as scribe, to Sandro] which is why in the email trail I suggested that the external-data-source version is clearer with a different construct 14:13:08 <AxelPolleres> +1 sandro. it is that only, not more, not less. 14:13:14 <sandro> Gary: All external means is you can't put it in the conclusion. 14:13:52 <AxelPolleres> ... thus we can allow any terms in External. is it up to the dialect author to define what that external schema complying to a specific term means. 14:14:08 <sandro> Gary: So if you want to say some term must never occur in a conclusion, then mark it external. 14:14:29 <AxelPolleres> ... so, I actualy have lost, WHAT we actually are discussing now? 14:14:44 <sandro> indeed. :) 14:14:49 <DaveReynolds> Paul: do we need to specify things external reference to Mismo in *RIF*, surely it is clear at runtime 14:16:26 <DaveReynolds> csma: but even if in production rule engines things are always external we should still mark that, to be conformant with other engines 14:16:49 <DaveReynolds> [scribe can't follow this well enough so nuances may be getting lost] 14:17:02 <DaveReynolds> Paul: can handle as a constraint in the spec 14:17:41 <DaveReynolds> csma: we need to differentiate between internal and external #/##, even if at the moment we only use external, but need distinction in future 14:17:43 <GaryHallmark> choice is External(Atom) or External(ATOMIC) -- can we just vote which we want??? 14:18:07 <DaveReynolds> Paul: at the moment no PR engine deals with non-external member/subclass 14:19:06 <DaveReynolds> Gary: choice is external(Atom) or external(ATOMIC) can we just vote on that then move on 14:19:31 <AxelPolleres> +1 for External( ATOMIC ) 14:20:04 <DaveReynolds> Paul: this is a usage issue, if PR engines only deal with external definitions, then is some future hypothetical case where you could also define internally really important to cover 14:20:11 <DaveReynolds> s/cover/cover?/ 14:20:38 <AxelPolleres> sorry, anybody else feeling that we waste valuable WG time here on a minor issue? 14:21:32 <sandro> I don't think it's minor, AxelPolleres, but I'm not sure this is the most productive approach to it. It seems important to understand externals, but I don't know how we'll get there. 14:21:33 <DaveReynolds> csma: if there are two cases we need two different syntaxes 14:22:19 <josb> I agree that we might want to cut short this discussion 14:22:29 <AxelPolleres> ... by reading the definition of exxternal schemata. 14:23:30 <DaveReynolds> Paul: using external to define a constraint on the implementation, isn't better to do that as a specification? 14:23:40 <DaveReynolds> s/better/it better/ 14:25:19 <DaveReynolds> Paul: could use test in PR and Logic system but in PR dialect could simply say always external without needed separate syntax, is that more flexible? 14:25:43 <DaveReynolds> csma: but in future dialects you could have dynamic membership, then you would lack a syntax 14:25:57 <DaveReynolds> Paul: but just define a new dialect PRDD that supports that new capability 14:26:19 <DaveReynolds> csma: BLD needs it now so why not define it now and use in PRD? 14:26:58 <DaveReynolds> Changhai: could use external predicate like instanceOf to represent these tests 14:27:25 <DaveReynolds> csma: would need to add isntanceOf etc to DTB for PRD 14:27:41 <DaveReynolds> s/isn/ins/ 14:28:26 <PaulVincent> Response to Axel: it could well be a minor issue - I've no idea whether LP guys will ever want to input production rules and do reasoning on the rules... (I guess this is the potential of a shared core) 14:28:51 <GaryHallmark> biz rule use case for membership in conclusion: if c#Customer and c[spent->$400] then c#GoldCustomer 14:28:53 <DaveReynolds> csma: not close to consensus? 14:29:23 <sandro> Chrisw: This issue is just External-Atom and External-Atomic 14:29:24 <Harold> BLD : 'External' '(' Atom | Frame ')' 14:29:31 <Harold> FLD : 'External' '(' ATOMIC ')') 14:29:31 <Harold> <Harold> ATOMIC ::= Const | Atom | Equal | Member | Subclass | Frame 14:30:43 <PaulVincent> I think Gary's example would translate to if ext(c#Customer) and ... 14:30:50 <sandro> csma: three ways: external(atom), external(atom|Frame), or external(atomic) 14:31:07 <DaveReynolds> Jos: this would require a new LC for BLD? 14:31:11 <AxelPolleres> Paul, but this discussion ONLY revolves around the allowed syntax to external calls, which is purely syntactic, the semantics of the External calls is not defined in the rule language, I could interpret an ?X#?Y as "?X goes well with ?Y and creamcheese" or whatever, upto the author defining that external schema. 14:32:09 <AxelPolleres> ... I don't see what this has to do with LP? 14:33:44 <PaulVincent> Axel: yes, but the issue is whether stuff that is ext will always be ext? ie C is a CreamChees may be ext in a PRD, but not in BLD - so immediately this makes sharing info more difficult? But, it might not be an issue... 14:33:51 <DaveReynolds> Straw test on choice: external(ATOM) external(ATOMIC) external([ATOM|Frame]) 14:34:37 <DaveReynolds> Straw test on choice: (1) external(ATOM) (2) external(ATOMIC) (3) external([ATOM|Frame]) 14:36:14 <AxelPolleres> Paul, redefining/reusing "#" in an external is as problematic/non-problematic as for instance defining an External( rif:iri( ...) ) predicate. 14:36:35 <AxelPolleres> ... where the latter is even possible with ATOM. 14:38:55 <josb> Why do people seem to assume that built-ins are the best way to access data sources? 14:39:02 <DaveReynolds> DaveReynolds has joined #rif 14:39:29 <DaveReynolds> Pref for (1) - 3 14:39:43 <DaveReynolds> Pref for (2) - 6 14:39:51 <ChrisW> (1) is EXTERNAL(ATOM) 14:40:02 <ChrisW> (2) is EXTERNAL(ATOMIC) 14:40:10 <josb> re 2: what does external equality mean? 14:40:26 <DaveReynolds> Pref for (3) - 2 14:40:56 <DaveReynolds> Object to (1) - 1 (MK) 14:41:01 <PaulVincent> What is Bob voting? :) 14:41:51 <DaveReynolds> Object to (2) - 1 (Jos) 14:41:59 <BobMoore> Bob isn't voting because he is struggling to figure out the consequences 14:42:21 <PaulVincent> Ah... like me 14:42:49 <BobMoore> I'm not really in favour of an option that means we HAVE to support dynamic class membership in PRD 14:42:52 <DaveReynolds> Object to (3) - 1 (Axel) 14:43:26 <BobMoore> Not that dynamic class membership is not a lot of fun when programming real world applications 14:43:35 <DaveReynolds> Jos - equality is equality, what makes external equality different from equality 14:44:00 <PaulVincent> Bob: of course, SRL does allow subclasses of externals to be defined in a ruleset. But I can't remember if you can change object membership at runtime... 14:44:25 <DaveReynolds> Axel - agrees with Gary's argument that frame but not #/## is strange 14:44:26 <BobMoore> ah but Nexpert allowed you to do that 14:45:07 <PaulVincent> Hmmm... but Nexpert was an internal object model and bwd chaining ... 14:46:55 <AxelPolleres> redefining/reusing "=" in an external schema is as problematic/non-problematic as for instance defining an External( rif:iri( ...) ) predicate. 14:47:54 <PaulVincent> Axel: I wonder if referencing an external nerual net / fuzzy logic system would be a use case for an ext(=) 14:49:41 <ChrisW> action: Chris to open an issue on what things are external 14:49:41 <trackbot> Created ACTION-601 - Open an issue on what things are external [on Christopher Welty - due 2008-10-04]. 14:50:23 <ChrisW> issue: Which to make external: ATOMIC, ATOM, or ATOM|FRAME 14:50:23 <trackbot> Created ISSUE-78 - Which to make external: ATOMIC, ATOM, or ATOM|FRAME ; please complete additional details at http://www.w3.org/2005/rules/wg/track/issues/78/edit . 14:50:28 <sandro> ACTION-601 closed 14:50:29 <trackbot> ACTION-601 Open an issue on what things are external closed 14:51:02 <ChrisW> rraagent, pointer? 14:51:09 <ChrisW> rrsagent, pointer? 14:51:09 <RRSAgent> See http://www.w3.org/2008/09/27-rif-irc#T14-51-09 14:52:01 <BobMoore> from a pragmatic perspecitve if I am building a rule or logic engine, everything is internal at "execution" time even if I have to "import" external definitions 14:52:07 <DaveReynolds> csma: so do not remove "at risk" on external frames but perhaps point to issue-78 in the text 14:52:35 <BobMoore> ... the problem arises when I try to map my internal representations back to an external representation 14:53:05 <PaulVincent> Bob: yes, its internal with constraints on what I can do... which is what I was commenting with Christian - PRDs using external schemas are just constrained on what they do... 14:53:47 <DaveReynolds> MK: have another option from Dave of adding another argument to external to specify the source see http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0193.html 14:54:23 <DaveReynolds> Next discussion: proposed modified text to BLD conformance clause 14:56:50 <DaveReynolds> "In addition, conformant BLD producers and consumers should preserve all annotations where possible" 14:57:39 <AxelPolleres> "In addition, conformant BLD producers and consumers SHOULD preserve all annotations" 14:57:57 <AxelPolleres> ... "where possible" is subsumed by should, isn't it? 15:00:05 <sandro> DaveReynolds: The "where possible" is good because it acknowledges that it may not always be possible. 15:01:01 <AxelPolleres> "In addition, conformant BLD producers and consumers External(should) preserve all annotations where possible OR 1=2." 15:01:24 <Harold> We could say: When you omit existing annotations, then you MUST add an annotation (* "annotation omitted" *)" at the enclosing syntactic construct. 15:02:12 <ChrisW> PROPOSED: remove round-tripping clause from conformance, add sentence to RIF_BLD clauses 15:02:19 <josb> +1 15:02:27 <Harold> +1 15:02:30 <AxelPolleres> +1 15:02:33 <AdrianP> +1 15:02:36 <Blaz> +1 15:02:37 <GaryHallmark> +1 15:02:41 <ChrisW> PROPOSED: remove round-tripping clause from BLD conformance, add sentence to RIF_BLD clauses 15:02:51 <ChrisW> PROPOSED: remove round-tripping clause from BLD conformance, add sentence to RIF_BLD clauses, closing round-tripping issue 15:03:08 <ChrisW> PROPOSED: remove round-tripping clause from BLD conformance, add sentence to RIF_BLD clauses, closing round-tripping issue 26 15:03:32 <DaveReynolds> No objections 15:03:33 <ChrisW> RESOLVED: remove round-tripping clause from BLD conformance, add sentence to RIF_BLD clauses, closing round-tripping issue 26 15:06:19 <ChrisW> rrsagent, pointer? 15:06:19 <RRSAgent> See http://www.w3.org/2008/09/27-rif-irc#T15-06-19 15:29:39 <Zakim> -BobMoore 15:32:35 <Zakim> -Meeting_Room 15:32:37 <Zakim> Team_(rif)13:34Z has ended 15:32:37 <Sandro> Topic: Parallel Session: PRD 15:32:39 <Zakim> Attendees were BobMoore, Meeting_Room 15:34:15 <PaulVincent> PaulVincent has joined #rif 15:41:31 <PaulVincent> Presume Bob will join the PRD conflict resolution discussion ... 15:42:04 <mdean> mdean has joined #rif 15:42:46 <BobMoore> Just dropped off the phone, will call back in when you restart 15:43:35 <DaveReynolds_> DaveReynolds_ has joined #rif 15:44:34 <AxelP> AxelP has joined #rif 15:45:16 <MichaelKifer> MichaelKifer has joined #rif 15:45:50 <PaulVincent> Paul to scribe... 15:46:00 <PaulVincent> Bob - we are restarting 15:46:23 <AdrianP> AdrianP has joined #rif 15:47:29 <Zakim> Team_(rif)13:34Z has now started 15:47:30 <saidtabet> saidtabet has joined #rif 15:47:36 <Zakim> +BobMoore 15:47:55 <csma> csma has joined #rif 15:47:56 <GaryHallmark> bob, slides are coming 15:47:59 <BobMoore> Waiting for the meeting room to dial back in 15:48:04 <GaryHallmark> and we will dial in soon 15:48:29 <PaulVincent> Bob - can you dial in now? 15:49:23 <BobMoore> I am already on the phone 15:49:39 <PaulVincent> CR = conflict resolution 15:50:16 <csma> zakim, what is the code? 15:50:16 <Zakim> the conference code is 26631 (tel:+1.617.761.6200 tel:+33.4.89.06.34.99 tel:+44.117.370.6152), csma 15:50:19 <Zakim> +??P1 15:51:02 <BobM> BobM has joined #rif 15:51:16 <PaulVincent> CSMA: details what is CR - which rule to fire when many can 15:52:00 <mdean> mdean has joined #rif 15:54:00 <PaulVincent> CSMA: CR covers all fireable instances of rules at a point in time including e.g. those handled by rule priority, also rules that may not be placed on agenda due to "no repeat" constraints 15:54:38 <PaulVincent> Gary: usually fire 1 rule, this may change WM, then other rules may not be fired 15:55:18 <PaulVincent> CSMA: yes ,we aretalking about rule execution cycle eg fire all actions without reinstantiating... 15:55:55 <PaulVincent> Gary: example of "fire all actions" could be a DB Trigger, but not commonly PR engines 15:56:49 <PaulVincent> Said: early CLIPS systems etc might execute all instance actions in same cycle 15:57:32 <AxelPolleres> AxelPolleres has joined #rif 15:57:34 <PaulVincent> CSMA: example of differentiation is sequential processing mode in BREs 15:57:55 <BobMoore> BobMoore has joined #rif 15:59:42 <BobMoore> BobMoore has joined #rif 16:02:03 <PaulVincent> Changhai: strategy could be set per rule 16:02:26 <ChrisW> FYI: FLD discussion is on the #fld channel 16:02:33 <PaulVincent> CSMA: Strategy defined per rule/ruleset, OR have a default 16:02:56 <PaulVincent> Said: multiple strategies is not practicable 16:03:51 <PaulVincent> CSMA: propose an annotation as default-for-all is unlikely; a standard set of annotations makes sense 16:04:38 <PaulVincent> Gary: lets discuss what the strategies are before discussing annotations 16:05:07 <PaulVincent> Said: also can have case of multiple cooperating engines 16:05:28 <PaulVincent> CSMA: OK so 3rd option is a programmable strategy 16:09:39 <PaulVincent> Changhai: practice might be a tag for strategy independent from execution mode 16:11:07 <PaulVincent> CSMA: Discussion on No Repeat / refraction / noloop - loops will continue to fire if the condition still holds 16:12:09 <PaulVincent> CSMA: eg if emp.salary < 10000 then emp.salary=emp.salary*1.1 ==> repeats until all emp have salary are >= 10000 16:15:32 <PaulVincent> CSMA: eg if alarm(X) then print alarm, alarm(1) and alarm(2), but on no-repeat print alarm only fires once 16:16:12 <PaulVincent> CSMA: no-repeat above applies on actions only in the same rule 16:16:55 <BobMoore> I'm quite happy with the concept of "no repeat" since it is the default behaviour of Blaze Advisor and I believe OPS5/Clips/Drools 16:18:06 <BobMoore> ... what maybe more of a problem is converting the description of the notion of "no repeat" into a more formal definition suitbable for the PRD specification 16:20:48 <BobMoore> Alarm example is weaker here than the salary example 16:21:39 <PaulVincent> CSMA: if alarm(X) and X.a>100 then print alarm(X) and no-repeat so will print once for alarm(1) and alarm(1).a>100 fact doesn't change 16:21:42 <BobMoore> Point in the salary example is you fire the rule to increase salary from 500 to 550, but not again to fire the rule from 550 to 605 16:22:41 <PaulVincent> Bob: votes for preferring Gary's example... 16:25:34 <PaulVincent> Gary: does issue change if there are intermediate changes? 16:26:12 <PaulVincent> CSMA: intermediate cycles have no effect (unless facts change) 16:29:37 <DaveReynolds_> DaveReynolds_ has joined #rif 16:32:31 <PaulVincent> Gary: JESS actions cannot add new instances during a rule action in the same rule cycle 16:33:56 <AxelPolleres> Dave, can you paste a link to the OWL Full/DL implementation you just mentioned? 16:36:37 <PaulVincent> Changhai: Rule agenda is key 16:36:40 <BobMoore> I think we are tiptoeing around the big problem which is that production rules are non-monatonic 16:37:28 <BobMoore> you can't understand what is going on without a veiw of the current state of working memory and the current state of the agenda 16:39:14 <PaulVincent> Said: qu is what you do to put rules on the agenda 16:40:57 <PaulVincent> CSMA: semantic description: we can consider that the agenda is always rebuilt, regardless of mechanisms to optimize this 16:42:01 <BobMoore> BobMoore has joined #rif 16:42:25 <AxelPolleres> BTW... the example misses External ( ... ) 16:43:54 <AxelPolleres> Do you agree that Exists ?x (And (ex:p(?x) pred:isNotInteger(?x) pred:isNotString(?x))) is true in RIF? 16:49:01 <AxelPolleres> ok, I overlooked/misread that. 16:51:29 <PaulVincent> Gary: need to define what is monitored for change in rules? 16:52:01 <PaulVincent> Axel: are you on the right IRC channel? 16:58:22 <PaulVincent> Paul: suggestion to check out all the various BRE semantics 16:58:35 <AxelPolleres> yes, I am 16:58:50 <AxelPolleres> ooops :-) 16:58:56 <AxelPolleres> on the wring 16:59:02 <PaulVincent> Changhai: ... but we can agree on refraction? 16:59:11 <AxelPolleres> s/ing/ong/ 17:00:45 <PaulVincent> CSMA: summary 1: no repeat: remove same instance as long as some condition holds 17:01:39 <PaulVincent> CSMA: 1b: what is "same instance"? 17:04:02 <PaulVincent> CSMA: 1b: ... binding variable in condition, or in action, or in both? 17:08:02 <PaulVincent> CSMA/Gary: looks at Example 1 again... 17:14:29 <PaulVincent> Paul: questions whether action references are indeed having any affect on agenda... 17:16:22 <PaulVincent> CSMA: proposal 1 doesnt work, possibly proposal 2 doesnt 17:16:33 <BobMoore> Paul - yes action references do have an effect on agenda 17:16:49 <BobMoore> ... or at least for some engines! 17:17:36 <PaulVincent> Bob - ues I recall SRL pattern - action can involve a fire-once response... 17:18:57 <PaulVincent> ,,, which is why we need to list examples to determine the subset for default in RIF... 17:19:10 <BobMoore> Paul - in SRL if you do not use a pattern (ie PRD variable) in the action the rule will only fire once. If you put in any action INCLUDING "ignore", the rule fires for all binding which satisfy the rule conditions 17:20:01 <PaulVincent> Bob - thx - so ignore is the BA switch for semantics... 17:20:33 <BobMoore> Paul - in effect yes 17:23:22 <PaulVincent> Bob - OK we need to include action references in the example list :) 17:24:31 <PaulVincent> CSMA: summary: we made some progress on mutual understanding 17:25:00 <BobMoore> Suggestion - we go back to the drawing board. and map out two or three scenario examples and those of us with a production rule engine say what the execution sequence will be for our engine and then compare notes to 17:25:08 <BobMoore> ... determine common ground 17:25:21 <PaulVincent> +1 17:25:40 <BobMoore> I think we are aguing about what PR system might do while for RIF we need to be clear about what real ones actually do 17:26:56 <PaulVincent> CSMA: Important: we need to define "keys" for strategies 17:27:08 <PaulVincent> closing discusssion 17:27:36 <PaulVincent> CSMA: correction: what are the variables for no repeat - but may only be UNIQUE keys 17:28:19 <PaulVincent> CSMA: reply to Bob - need common ground, agree 17:29:09 <Zakim> -??P1 17:29:13 <Zakim> -BobMoore 17:29:15 <Zakim> Team_(rif)13:34Z has ended 17:29:15 <Zakim> Attendees were BobMoore 17:29:36 <BobMoore> when will you be back from lunch? 18:17:49 <DaveReynolds_> DaveReynolds_ has joined #rif 18:22:19 <csma> We are back 18:22:30 <csma> not started yet, though 18:22:34 <csma> I will ping you 18:26:34 <josb> josb has joined #rif 18:33:10 <csma> Bob, we start (Actions in PRD) 18:33:23 <DaveReynolds_> DaveReynolds_ has joined #rif 18:33:57 <csma> We are dialing right now 18:34:36 <Zakim> Team_(rif)13:34Z has now started 18:34:40 <GaryHallmark> zakim, what is the code? 18:34:40 <Zakim> the conference code is 26631 (tel:+1.617.761.6200 tel:+33.4.89.06.34.99 tel:+44.117.370.6152), GaryHallmark 18:34:43 <Zakim> +??P0 18:34:56 <saidtabet> saidtabet has joined #rif 18:35:07 <PaulVincent> PaulVincent has joined #rif 18:35:31 <PaulVincent> Bob - we are reconvening FYI 18:37:21 <GaryHallmark> ACTION ::= ASSERT | Retract | New | ACTION_BLOCK 18:37:23 <GaryHallmark> ASSERT ::= Atom | Frame 18:37:24 <GaryHallmark> New ::= 'New' ( Var '#' )? Const 18:37:26 <GaryHallmark> Retract ::= 'Retract' '(' Atom | Frame | TERM ')' 18:37:27 <GaryHallmark> ACTION_BLOCK ::= 'And' '(' ASSERT+ ')' | 'Do' Var* '(' ACTION* ')' 18:37:36 <csma> zakim, ??P0 is MeetingRoom 18:37:37 <Zakim> +MeetingRoom; got it 18:38:15 <GaryHallmark> rrsagent, delete action 30 18:38:15 <RRSAgent> I'm logging. I don't understand 'delete action 30', GaryHallmark. Try /msg RRSAgent help 18:38:20 <AdrianP> AdrianP has joined #rif 18:38:38 <Blaz> Blaz has joined #rif 18:40:36 <AdrianP> AdrianP has joined #rif 18:41:02 <AdrianP> Zakim, who is on the phone? 18:41:02 <Zakim> On the phone I see MeetingRoom 18:41:12 <PaulVincent> Gary: assert atom = add a tuple, assert a frame = add a slot, and inverse for retract; retract term is remove all slots and membership 18:41:50 <MichaelKifer> MichaelKifer has joined #rif 18:42:11 <MichaelKifer_> MichaelKifer_ has joined #rif 18:42:17 <PaulVincent> Gary: new creates a new obj and adds onjid into membership for given class 18:42:19 <MichaelKifer_> #fld 18:42:35 <Harold> Harold has joined #rif 18:43:26 <PaulVincent> CSMA: constructor gives default values means this must be explicitly defined in action part 18:44:44 <PaulVincent> http://www.w3.org/2005/rules/wiki/PRD#Presentation_syntax 18:45:28 <BobMoore> Dialing in now - I can only stay for a short while as it is family supper time soon 18:45:58 <Zakim> +BobMoore 18:46:25 <PaulVincent> Gary: Looking at example 2.11 18:47:24 <MichaelKifer_> MichaelKifer_ has joined #rif 18:48:40 <PaulVincent> CSMA: why not allow initialization with values? 18:50:08 <PaulVincent> Gary: in OBR we do Java new + setProperty calls 18:53:22 <PaulVincent> Gary: multiple "new"s require multiple variables 18:53:38 <GaryHallmark> GaryHallmark has joined #rif 18:56:19 <PaulVincent> Gary: reviews draft semantics in 3.3 19:00:31 <GaryHallmark> GaryHallmark has joined #rif 19:02:24 <Zakim> -BobMoore 19:02:25 <BobMoore> Supper time - I'll be back in 30 mins or so 19:03:34 <PaulVincent> Gary: working memory union with asserted atom a may need to be defined as not a set union but mappings to the domain 19:04:05 <csma> PROPOSED: add construct DO in then-part of PRD rules to declare ?local variables for New frames (according to Gary's proposal) 19:04:15 <PaulVincent> Harold: clarify this is just the action part of a rule 19:04:35 <csma> PROPOSED: add construct New with Gary's proposed syntax and semantics in then-part of PRD rules 19:05:55 <GaryHallmark_> GaryHallmark_ has joined #rif 19:07:18 <csma> PROPOSED: add possibility to Retract a TERM in then-part of PRD rules with Gary's proposed syntax, to remove a frame object 19:08:43 <PaulVincent> Paul: issue is with external constructors and mapping from multiple asserts to that 19:12:00 <PaulVincent> CSMA: methods needed too per yesterday discussion 19:12:53 <PaulVincent> CSMA: no time to discuss modify semantics 19:14:03 <GaryHallmark_> GaryHallmark_ has joined #rif 19:15:12 <csma> PROPOSED: add possibility to Retract a TERM in then-part of PRD rules with Gary's proposed syntax, to remove a frame object, that is, remove any reference to that object in the working memory 19:17:07 <Zakim> -MeetingRoom 19:17:08 <Zakim> Team_(rif)13:34Z has ended 19:17:08 <Zakim> Attendees were MeetingRoom, BobMoore 19:17:26 <csma> PROPOSED: add possibility to Retract a TERM in then-part of PRD rules with Gary's proposed syntax, to remove a frame object, that is, remove object from the instances of its class as well as all the frames with that object 19:18:11 <PaulVincent> Gary: can't use ext constructors as these are methods (could be added later, but need to handle asserts now) 19:19:12 <csma> PROPOSED: add possibility to Retract a TERM in then-part of PRD rules with Gary's proposed syntax, to remove a frame object, that is, remove object from the instances of its class as well as all the frames with that object in the object position 19:22:20 <csma> PROPOSED: add construct DO in then-part of PRD rules to declare ?local variables for New frames (according to Gary's proposal) 19:22:24 <PaulVincent> Adrian: would like to re-use Do for local vars for other uses 19:22:33 <PaulVincent> CSMA: ... but this would be a separate different proposal 19:23:45 <csma> PROPOSED: add construct DO in then-part of PRD rules to declare ?local variables for New frames (according to Gary's proposal); not excluding that later resolutions might extend the use of local variables in the action part. 19:24:57 <PaulVincent> Paul: concern that new(obj) --> pattern of rif statements --> new(obj) is an overload 19:25:15 <PaulVincent> CSMA: later we could use aliases for that, but want to retain compatibility 19:26:15 <csma> PROPOSED: add construct New with Gary's proposed syntax and semantics in then-part of PRD rules 19:29:07 <csma> PROPOSED: add construct New with Gary's proposed syntax and semantics in then-part of PRD rules; not excluding extending the use of constructors once we resolve how to call "methods". 19:31:19 <csma> PROPOSED: add construct New with Gary's proposed syntax and semantics in then-part of PRD rules; not excluding extending it later with the use of constructors once we resolve how to call "methods" 15:32:37 <Sandro> Topic: Parallel Session: FLD, DTB 15:32:37 <Sandro> subtopic: FLD 15:43:22 <RRSAgent> RRSAgent has joined #fld 15:43:22 <RRSAgent> logging to http://www.w3.org/2008/09/27-fld-irc 15:53:33 <ChrisW> FLD open items 15:55:07 <ChrisW> add aggregates 15:59:41 <Blaz> Blaz has joined #fld 15:59:44 <josb> josb has joined #fld 15:59:49 <DaveReynolds> DaveReynolds has joined #fld 16:00:07 <mdean> mdean has joined #fld 16:00:40 <Harold> Harold has joined #fld 16:00:56 <ChrisW> 1) modules 16:00:59 <ChrisW> 2) aggregates 16:01:54 <Harold> 3) external (contingent on what we need in BLD) 16:02:12 <AxelPolleres> AxelPolleres has joined #fld 16:06:59 <AxelPolleres> Why does PRD use NmNot for Naf? 16:07:21 <AxelPolleres> I think it should be fine if they use Naf. 16:08:36 <AxelPolleres> If we talk about Naf in the body, what about Naf in the haed, Disjunction in the head? 16:09:25 <Harold> "Equivalence of well-founded and stable semantics" (Francoise Gire): 16:09:27 <Harold> http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.47.3980 16:14:30 <Harold> Patricia Hill, John Lloyd: Gödel Programming Language (http://www.cs.bris.ac.uk/~bowers/goedel.html). 16:15:50 <ChrisW> Posible dialect (for testing FLD, as well as being potentially useful): 16:16:14 <ChrisW> locally stratified neg + removing "unique context" requirement on contstants 16:29:39 <DaveReynolds_> DaveReynolds_ has joined #fld 16:38:14 <ChrisW> Proposal: have three dialects: BLD w/o negative guards, BLD w/o equality in the head, and BLD 17:00:15 <AxelPolleres> Exists ?X (http://example.com/qqq"^^rif:iri(?X) ) would be entailed, yes? 17:00:43 <AxelPolleres> ... in the example of sec 3.9. 17:04:55 <AxelPolleres> So, the answer to the query Exists ?X (http://example.com/qqq"^^rif:iri(?X) ) would be true, but there is no concrete instance for ?X which can be output as answer. 18:17:50 <DaveReynolds_> DaveReynolds_ has joined #fld 18:26:35 <josb> josb has joined #fld 18:33:24 <DaveReynolds_> DaveReynolds_ has joined #fld 18:38:38 <Blaz> Blaz has joined #fld 18:41:25 <sandro> sandro has joined #fld 18:41:26 <AxelPolleres> AxelPolleres has joined #fld 18:41:36 <sandro> RRSAgent, pointer? 18:41:36 <RRSAgent> See http://www.w3.org/2008/09/27-fld-irc#T18-41-36 18:41:45 <trackbot> trackbot has joined #fld 18:41:45 <trackbot> Sorry... I don't know anything about this channel 18:41:45 <trackbot> If you want to associate this channel with an existing Tracker, please say 'trackbot, associate this channel with #channel' (where #channel is the name of default channel for the group) 18:42:04 <sandro> trackbot, associate this channel with #rif 18:42:04 <trackbot> Associating this channel with #rif... 18:42:15 <sandro> issue-79? 18:42:15 <trackbot> ISSUE-79 does not exists 18:42:20 <sandro> issue-78? 18:42:20 <trackbot> ISSUE-78 -- Which to make external: ATOMIC, ATOM, or ATOM|FRAME -- OPEN 18:42:20 <trackbot> http://www.w3.org/2005/rules/wg/track/issues/78 18:43:02 <ChrisW> action: michael to update entailment definition to be for documents and revert rif:local semantics 18:43:02 <trackbot> Sorry, amibiguous username (more than one match) - michael 18:43:02 <trackbot> Try using a different identifier, such as family name or username (eg. mkifer, msintek, merdmann) 18:43:11 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2008Aug/0074.html 18:43:13 <ChrisW> action: mkifer to update entailment definition to be for documents and revert rif:local semantics 18:43:14 <trackbot> Created ACTION-602 - Update entailment definition to be for documents and revert rif:local semantics [on Michael Kifer - due 2008-10-04]. 18:43:29 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0000.html 18:43:34 <sandro> jos: entailment will be for both documents and conditions. 18:43:47 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0142.html 18:44:32 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0187.html 18:44:46 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0194.html 18:45:48 <ChrisW> issue: negative guards in DTB - is this another dialect? 18:45:48 <trackbot> Created ISSUE-79 - Negative guards in DTB - is this another dialect? ; please complete additional details at http://www.w3.org/2005/rules/wg/track/issues/79/edit . 18:48:10 <MichaelKifer_> MichaelKifer_ has joined #fld 18:52:33 <sandro> subtopic: issue-67 string-less-than 18:53:17 <sandro> chrisw: who wants only fn:compare? 18:53:28 <sandro> 1 18:53:32 <sandro> er 2 18:53:53 <sandro> chrisw: who wants the added stuff? 18:53:56 <sandro> 5 or so. 18:58:36 <sandro> chrisw: we'll have to take this back to the main group. 18:58:50 <sandro> subtopic: issue-61 Casting to/from rif:iri 18:59:01 <sandro> All DTB issues: http://www.w3.org/2005/rules/wg/track/products/18 19:01:57 <sandro> chrisw: this should have beenn closed as part of resolution at some past meeting. 19:04:57 <sandro> subtopic: operators on rif:text 19:05:19 <sandro> DaveReynolds: I'd like equal, but not other compare....? 19:06:03 <AxelPolleres> which functions should we have for rdf:text? 19:09:17 <AxelPolleres> equal inequality might be useful. 19:10:18 <sandro> Dave: Just define equality as equality of the lexical form. 19:11:00 <sandro> ACTION: axel to discuss in rdf:text task force the including of certain functions and preficates, and name of extractor (eg func:lang-from-text) 19:11:00 <trackbot> Created ACTION-603 - Discuss in rdf:text task force the including of certain functions and preficates, and name of extractor (eg func:lang-from-text) [on Axel Polleres - due 2008-10-04]. 19:13:29 <sandro> Dave: Just use string-equality on the lexical represtnation of rdf:XMLLiteral. It's just sugar for compare of string version (assuming there is one). 19:13:49 <sandro> Axel: These is no cast from rdf:XMLLiteral to string. 19:13:53 <sandro> Jos: Let's add one 19:13:59 <sandro> +1 19:14:07 <sandro> DaveReynolds: +1 19:14:58 <sandro> Axel: cast string to XMLLiteral, as well? 19:15:02 <sandro> Sandro: I'd think so. 19:15:47 <AxelPolleres> XMLLiteral to and from string conversions (casts) plus equals/not-equals 19:16:48 <sandro> ACTION: Axel add text for casting to string and equal/not-equal, an (for discussion) from-string (which requires implementing XML canonicalization. 19:16:48 <trackbot> Created ACTION-604 - Add text for casting to string and equal/not-equal, an (for discussion) from-string (which requires implementing XML canonicalization. [on Axel Polleres - due 2008-10-04]. 19:17:19 <sandro> Dave: Fiddling with XML at the string level isn't very useful --- we'd like nice XML support some day. 19:20:19 <AxelPolleres> 13) Editor's Note: No less-than-or-equal or greater-than-or-equal predicates are defined in this draft for durations, since there are no separate op:dayTimeDuration-equal nor op:yearMonthDuration-equalpredicates in [XPath-Functions], but only a common predicate op:duration-equal. Future versions of this working draft may resolve this by introducing new equality predicates pred:dayTimeDuration-equal and pred:yearMonthDuration-equal with restricted inten 19:20:19 <AxelPolleres> PROPOSED: introduce a single predicate duration-equal that only evaluates to true if the arguments are both of the same duration subtype and equal. 19:25:41 <sandro> Chrisw: How did we end up with BLD forcing only one arity per function/predicate? 19:26:29 <sandro> Jos: There are two different ways to handle the semantics of this, and each had problems, and we wanted to be able to extend to FOL. 19:26:36 <sandro> Chrisw: right, okay. 19:29:33 <sandro> Axel: we've defined strings to be rdf:text's with an empty language tag. 19:31:00 <sandro> Chrisw: Don't do that! I don't want every xs:string to also be an rdf:text. 19:31:20 <DaveReynolds> http://www.w3.org/2007/OWL/wiki/InternationalizedStringSpec#xsd:string_as_a_restriction_of_rdf:text 19:44:02 <sandro> Chrisw: rdf:text needs two builtins, one to get the string, one to get the language tag. And later we figure out what to do with casting. 19:44:09 <sandro> RRSAgent, make minutes 19:44:09 <RRSAgent> I have made the request to generate http://www.w3.org/2008/09/27-fld-minutes.html sandro 19:45:19 <Sandro> Topic: Debrief from Parallel Sessions 19:45:19 <DaveReynolds_> DaveReynolds_ has joined #rif 19:45:33 <Zakim> Team_(rif)13:34Z has now started 19:45:40 <Zakim> +BobMoore 19:46:32 <BobMoore> can the meeting room rejoin? 19:47:37 <DaveReynolds> Bob - doing it now 19:48:29 <Zakim> +??P1 19:49:15 <Harold> Debrief 19:49:26 <sandro> subtopic: Debrief of "FLD" Breakout 19:49:38 <DaveReynolds> zakim, ??P1 is Meeting_Room 19:49:38 <Zakim> +Meeting_Room; got it 19:50:23 <josb> another nasty test case: http://www.w3.org/2005/rules/wiki/Disjunctive_Information_from_Unsafe_Builtins 19:50:31 <Harold> Chrisw: FLD should have Modules (Michael). 19:51:39 <Harold> Then discussed BLD. 19:52:55 <Harold> Based on Jos' test cases that use negative guards, eq in head, and sneak in disjunction. 19:54:51 <Harold> subtopic: Debrief of "DTB" Breakout 19:56:03 <Harold> Chrisw: String compare operators from XPath are not symmetric between strings and numbers. 19:57:36 <Harold> Issue 61 (casting to/from IRIs) should be formally closed, since it was de facto already in a telecon. 19:58:35 <ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2008Aug/att-0012/2008-08-05-rif-minutes.html 19:58:45 <sandro> PROPOSED: Close issue-61, since the underlying decision was already settled in 08-05 meeting 19:59:01 <sandro> RESOLVED: Close issue-61, since the underlying decision was already settled in 08-05 meeting 19:59:07 <sandro> issue-61 closed 19:59:22 <sandro> action: chris close issue 61 19:59:22 <trackbot> Created ACTION-605 - Close issue 61 [on Christopher Welty - due 2008-10-04]. 19:59:47 <DaveReynolds> s/RESOLEVED/RESOLVED/ 19:59:56 <Harold> Discussed rdf:text (int'lized string). 20:00:16 <Harold> Should bring in DTB built-ins. 20:00:59 <sandro> Chrisw: we agreed rdf:text should have accessor functions for its two components. casting is not resolved. 20:01:05 <Harold> Casting of rdf:text discussed: string and lang tag. 20:02:07 <Harold> We need to decide on fn:compare. 20:02:28 <sandro> would look like: PROPOSED: Add fn;string-less-than, etc, in addition to fn:compare, to make string comparison symmetric with numeric comparison. 20:02:34 <Harold> Also, remaining editor notes. 20:02:55 <Harold> Dave: Did you discuss subdialects. 20:04:36 <Harold> Chrisw: One possibility: 3 dialects, taking out equality in head, negative guards, or both. 20:04:47 <Harold> (from BLD.) 20:06:08 <Harold> subtopic: Debrief of "PRD" Breakout 20:07:01 <Harold> Christian: Slides on Conflict Resolution 20:09:47 <Harold> "Same instance" and "Same binding" discussions. 20:10:46 <Harold> Nothing to propose here at this point. 20:11:02 <Harold> Christian: Actions in PRD 20:11:31 <StellaMitchell> StellaMitchell has joined #rif 20:11:58 <Harold> PROPOSED: 'New' construct for action part of PRD rules. 20:12:48 <Harold> PROPOSED: 'Do' for action part of PRD rules. 20:13:22 <Harold> Makes specified variables local to its scope. 20:14:19 <Harold> PROPOSED: 'Retract' for action part of PRD rules extended from atoms and frames to TERMs. 20:18:09 <Harold> The TERM-identified frame information is removed: their object from the Member relation and the frames describing this object. 20:18:59 <Harold> Michael: why not use Erase? 20:20:06 <Harold> Jos: TERM is ambiguous here. 20:20:20 <Harold> Discussion: doesn't seem so. 20:20:53 <Harold> Christian: Can we pass these resolutions. 20:21:10 <Harold> Dave: Two of the three. 20:21:33 <Blaz> Blaz has joined #rif 20:21:40 <Harold> Michael: What about obj[], ie nullary? 20:23:27 <Harold> Dave/Michael/Christian: Since there are no function symbols in Core there is no ambiguity. 20:24:16 <Harold> Michael: Yet, a new name (Erase?) would be good. 20:25:20 <Harold> Michael: What is 'Do' ? 20:25:43 <Harold> Gary: Corresponds to 'Exists' in BLD. 20:25:45 <AxelPolleres> AxelPolleres has joined #rif 20:25:47 <Harold> ... W 20:26:15 <Harold> e need to clarify single-assignment vs. re-assignment. 20:26:36 <Harold> Further discussion in coming Taskforce telecon. 20:27:20 <Harold> topic: Process wrt f2f minutes 20:27:46 <Harold> Sandro: Gets better and better. 20:28:14 <Harold> Cool stuff like moves all Resolutions to the top. 20:29:11 <Harold> Easier to fix typos, inaccurate renderings of what people said, etc. 20:29:41 <Harold> (especially for scribes) 20:30:40 <Harold> When command Scribenick was forgotten, can still be easily fixed now. 20:31:37 <Harold> Christian: The way the history is changed is visible from the original archive. 20:32:03 <Harold> ... Benefit of wiki: everyone can edit. 20:32:37 <Harold> Sandro: What about: If you edit something in the history , send email to the scribe. 20:33:16 <Harold> Point them to the diff of your changes. 20:34:16 <Harold> There is also topic:, subtopic: subsubtopic: 20:34:31 <Harold> Creates indentation in toc. 20:42:25 <Blaz> Blaz has joined #rif 20:58:12 <AdrianP> AdrianP has joined #rif 20:58:29 <sandro> chrisw: how close are we to next WD of PRD? 20:58:38 <AdrianP> scribenick AdrianP 20:58:38 <Sandro> topic: Work Planning 20:58:48 <sandro> csma: we have three items we're working on..... 20:59:16 <AdrianP> Christian: close on the action side as next step 20:59:30 <AdrianP> reasonable close on the semantics 21:01:17 <AdrianP> csma: conflict resolution progress in next 2 weeks 21:01:43 <AdrianP> csma: frozen version of PRD by October 28th 21:02:41 <AdrianP> csma: review two weeks after in Nov. 21:03:01 <AdrianP> axel: same schedule for DTB 21:04:32 <AdrianP> adrian: action was to add the test cases to working draft 21:04:42 <AdrianP> stella: need some feedback from the grouü 21:04:54 <AdrianP> s/grouü/group 21:05:25 <AdrianP> adrian same schedule for Test Cases; in two weeks 21:05:34 <AdrianP> chrisw: two weeks from now we freeze 21:07:16 <AdrianP> michael: FLD open issues are aggregates, modules, .. 21:07:59 <AdrianP> michael: October, 28th for frozen version 21:09:00 <AdrianP> chrisw: any open issues for BLD? 21:09:49 <AdrianP> chrisw: nothing changes; some clarifications wrt entailment 21:10:03 <AdrianP> michael: have to fix defitions 21:10:16 <AdrianP> josb: how are changes documented? 21:11:15 <AdrianP> harold: could we point to the diff from WIKI? 21:12:05 <sandro> ACTION: MichaelKifer add "Changes Since Last Call" section to BLD 21:12:05 <trackbot> Sorry, couldn't find user - MichaelKifer 21:12:12 <sandro> ACTION: Kifer add "Changes Since Last Call" section to BLD 21:12:12 <trackbot> Created ACTION-606 - Add \"Changes Since Last Call\" section to BLD [on Michael Kifer - due 2008-10-04]. 21:12:46 <sandro> chrisw: pure grammar/spelling changes + clarificaitons + more detailed bugfix change. 21:14:32 <AdrianP> sandro: new section "change log" 21:14:51 <Harold> After: 10 Appendix: RIF Media Type Registration 21:15:07 <Harold> 10 Appendix: change log 21:15:08 <AdrianP> jos: SWC; fixed an error 21:15:10 <sandro> ACTION: Jos to add change log to SWC 21:15:10 <trackbot> Sorry, amibiguous username (more than one match) - Jos 21:15:10 <trackbot> Try using a different identifier, such as family name or username (eg. jdebruij2, jderoo) 21:15:16 <sandro> ACTION: Josb to add change log to SWC 21:15:16 <trackbot> Created ACTION-607 - Add change log to SWC [on Jos de Bruijn - due 2008-10-04]. 21:15:27 <AdrianP> jos: end of October as deadline for SWC 21:15:40 <Harold> 11 Appendix: Changes since Last Call 21:16:17 <sandro> Chrisw: CR? 21:17:00 <AdrianP> chrisw: candidate recommendation Nov. 14th 21:20:07 <AdrianP> Sandro: DTB has to CR at the same time 21:20:19 <AxelPolleres> negative guards were introduced for modeling if-then-else behavior in terms of "type-checking", which you can no longer do really if you restrict the domain for negative guards. 21:20:50 <AdrianP> Sandro: we need to get DTB to last call before BLD CR 21:21:03 <AxelPolleres> ... so, basically, before restricting the intended domain for neg guards, I'd rather drop them entirely. 21:21:31 <AdrianP> Sandro: ... maybe "ongoing last call for DTB" will work 21:21:58 <AdrianP> Chrisw: Core? 21:22:13 <DaveReynolds> Axel - an example would be the OWL RL rules, for literal values need type checking rules which need to negative guards but only over primitive datatypes 21:23:12 <AxelPolleres> Dave - you don't need "real" negation semantics there is what you say? 21:23:18 <josb> Also the embedding of OWL DLP combinations requires the negative guards with the current semantics 21:23:19 <AdrianP> Chrisw: UCR? 21:23:20 <josb> http://www.w3.org/2005/rules/wiki/SWC#Embedding_OWL_DLP_into_RIF_BLD 21:23:31 <AdrianP> Chrisw: Same schedule for UCR as for other documents 21:23:42 <AdrianP> Chrisw: next f2f 21:23:53 <AdrianP> csma: December in Portland? 21:24:08 <josb> w/o "real" negative guards, I don't think OWL DLP combinations can be embedded 21:24:34 <AdrianP> chrisw: January Portland? 21:25:34 <sandro> ACTION: Gary to tell Sandro what dates work for F2F12. 21:25:34 <trackbot> Created ACTION-608 - Tell Sandro what dates work for F2F12. [on Gary Hallmark - due 2008-10-04]. 21:25:36 <AdrianP> Gary: will send possible dates 21:25:46 <AdrianP> Chrisw: Core? 21:25:54 <AxelPolleres> ... yup jos, what I tried to say was: restricting the domain doesn't seem to let us end up in something very useful, does it? 21:25:57 <Harold> http://lists.w3.org/Archives/Public/public-rif-wg/2008Sep/0151.html 21:25:58 <sandro> topic: Core 21:25:59 <AdrianP> Harold: sent extract from Monday telecon 21:26:11 <josb> not so useful, it seems 21:27:36 <AdrianP> Harold: discussions from this telecon need to be discussed furhter 21:29:02 <AdrianP> Chrisw: Can we close issues? 21:29:39 <AdrianP> mk: equality can not be on data types 21:31:59 <AdrianP> christian: does unrestricted equality make sense in PRD? 21:32:38 <AdrianP> ´PROPOSED: Core should keep unrestricted equality and external function and predicate calls in rule bodies and keep external functions calls in rule heads. 21:33:35 <sandro> PROPOSED: Core should keep unrestricted equality and external function and predicate calls in rule conditions and keep external functions calls in rule conclusions 21:33:53 <sandro> csma: +0 21:34:16 <sandro> RESOLVED: Core should keep unrestricted equality and external function and predicate calls in rule conditions and keep external functions calls in rule conclusions 21:34:42 <sandro> http://www.w3.org/2005/rules/wg/track/issues/71 21:35:36 <csma> My question was whether we will still want unrestricted equality in conditions in PRD, if we are allowed externally specified equality... 21:35:57 <sandro> (That closes issue-76, Equality in Core) 21:36:20 <sandro> ACTION: Chris to close issue-76 21:36:20 <trackbot> Created ACTION-609 - Close issue-76 [on Christopher Welty - due 2008-10-04]. 21:36:49 <sandro> PROPOSED: close issue-71, given we're retaining equality in conditions in core 21:37:00 <sandro> RESOLVED: close issue-71, given we're retaining equality in conditions in core 21:37:03 <AdrianP> chrisw: closed issue 71 and 76 21:37:04 <sandro> ACTION: Chris to close issue-71 21:37:04 <trackbot> Created ACTION-610 - Close issue-71 [on Christopher Welty - due 2008-10-04]. 21:38:13 <Zakim> -BobMoore 21:38:28 <AdrianP> PROPOSED: Core should keep both frames/objects and (positional-argument) predicates/relations. 21:39:34 <sandro> PROPOSED: Close issue-74, saying Core will have both Predicates (with positional arguments) and Frames (no comment on having membership) 21:39:43 <sandro> RESOLVED: Close issue-74, saying Core will have both Predicates (with positional arguments) and Frames (no comment on having membership) 21:40:04 <sandro> action: chris closs issue-74 21:40:04 <trackbot> Created ACTION-611 - Closs issue-74 [on Christopher Welty - due 2008-10-04]. 21:40:31 <AdrianP> PROPOSED: Core should keep disjunction in rule bodies, only if this is permitted by the solution to issue-70. 21:40:35 <AdrianP> Jos: Why? 21:40:56 <sandro> Jos: Why have disjunction in rule bodies? 21:41:00 <AdrianP> Gary: disj. often used in business rules 21:41:07 <sandro> Gary: It's frequently used in real rules. 21:41:41 <AdrianP> Gary: counterproposal is two write a huge number of rules 21:42:01 <AdrianP> Jos: now you put the burden on rule engine vendors without disjunction 21:42:25 <AdrianP> csma: question is should Core be the largest common subset between PR and logic languages 21:42:38 <AdrianP> dave: we agreed smaller then that 21:43:11 <AdrianP> csma: burden is always somewhere 21:43:33 <AdrianP> csma: nothing changes for PRD 21:43:58 <AdrianP> Gary: can interchange between logic and production language using Core 21:45:04 <AxelPolleres> I reckon there is not necesarily a blowup, you can introduce new symbols in the rewriting of discjuntions, the rewriting would not be exponential. 21:45:11 <AdrianP> Sandro: you don't want to send out a huge RIF document if you can avoid it 21:45:20 <AxelPolleres> Agreed? 21:45:43 <AdrianP> Paul: whole idea is interchange 21:46:02 <AdrianP> csma: I do not go through Core, but use directly PRD 21:46:14 <AdrianP> csma: question is are there users which make use of Core 21:46:48 <AdrianP> gary: would like to represent rules in smallest possible dialect 21:47:01 <AdrianP> chrisw: sounds like we can not close issue today 21:47:23 <AdrianP> harold: was consensus in the task force 21:48:04 <AdrianP> gary: my system can handle disjunctions with bound variables 21:48:26 <AdrianP> axel: some people need to rewrite in non disjunctive rules 21:49:43 <AdrianP> jos: the way it is phrased there are no restrictions on the disjunction 21:49:55 <AdrianP> csma: it's related to issue 70 21:50:45 <AdrianP> jos: nesting of disjunction is the problem 21:51:01 <AdrianP> michael: you either get it or have to explode by your self 21:51:19 <AdrianP> gary: better in the translator than having large rule sets 21:51:29 <AdrianP> chrisw: postpone it to next telecon 21:52:34 <AdrianP> chrisw: other business? 21:52:38 <Harold> http://www.w3.org/2005/rules/wiki/OWLRL 21:52:47 <AdrianP> dave: take a look at OWL RL 21:53:15 <AdrianP> axel: have to look into our guard solutions 21:55:01 <AdrianP> dave: translate into RIF Core rules, we support different datatypes, 21:56:27 <AdrianP> dave: we can implement translator from OWL RL into RIF 21:56:56 <sandro> ADJOURN </nowiki></pre>

• Template:Chatlog

Return to Chatlog 2008-09-27.