Admin

<csma> http://lists.w3.org/Archives/Public/public-rif-wg/2009Feb/att-0063/10-Feb-2009-rif-mins.html

<csma> PROPOSED: accept the minutes of February 10

<csma> RESOLVED: Accept the minutes of February 10

csma: did not resolve datetimestamp issue last time. tabled waiting for Dave's input.

Liason

jos: rif/owl summary - owl still likes non-disjointness
... Alan is the holdout

chris: most scientists and engineers agree about disjointness. 0 != .000

sandro: disjointness on agenda for owl f2f next week

csma: DTB LC blocked on disjointness issue

<josb> my sense is we need at *least* 1 month for DTB LC

F2F13

csma: looks like f2f13 preferences leaning toward Boston

<sandro> csma: F2F13 http://www.w3.org/2002/09/wbs/38457/f2f13_dates/results

<ChrisW> I prefer tues-thurs to allow travel to spain

<josb> I prefer Wed-Fri

<sandro> http://www.w3.org/2002/09/wbs/38457/f2f13_dates/results#xq3

<gary> W-F is best for me

<josb> just pref.

<AdrianP> preference for me to stay over Saturday

<AdrianP> as it usually much cheaper to fly

<AdrianP> WWW09 starts on Monday

<ChrisW> michael?

<csma> Micheal, will you able to attend F2F13 on Friday 17 April?

<sandro> We're trying to decide between 14-16 April and 15-17 April.

<Michael_Kifer> yes

<ChrisW> (what question was that an answer to?)

summary: Boston April 15-17 seems best

<csma> next item

issue-90

<csma> PROPOSED: RIF-RDF combinations are not defined in case rif:iri or rdf:text are used in the imported RDF graphs

<csma> RESOLVED: RIF-RDF combinations are not defined in case rif:iri or rdf:text are used in the imported RDF graphs

<sandro> issue-90?

<trackbot> ISSUE-90 -- Can rif:iri and rdf:text be used as datatypes in RDF graphs, in combination with RIF? ie approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_1, http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Enta -- OPEN

<trackbot> http://www.w3.org/2005/rules/wg/track/issues/90

<josb> remove

sandro: let's keep the tests as some kind of negative test

<sandro> sandro: "import error test" or some such

chris: "undefined" is not an error, we aren't telling what it means
... RIF doesn't tell you what it means
... negative entailment is too strong

csma: how are other undefined things handled?
... e.g. w.r.t. externals

chris: neg. entailment is an attempt at providing a defn, but we are talking about undefined

jos: can't write a test case about an undefined thing

<sandro> jos: If you import an RDF graph that contains rif:iri or rdf:text as a datatype, then the entire combination is simply undefined.

several people: saying its an error is about the same as saying it is undefined

<sandro> 'because these are undefined, implementations SHOULD provide a warning or run-time error in such event'

chris: implementations SHOULD give warning when triples containing rdf:text or rif:iri are imported

<josb> http://www.w3.org/2005/rules/wiki/SWC#Interpretation_of_Profiles

<sandro> jos: we already have a bit about "rejecting document", so include this as a reject situation.

<sandro> sandro: we could make this an Import Rejection test case

<ChrisW> ACTION: josb to add text to SWC indicating "implementations should reject" graphs with rdf:text etc [recorded in http://www.w3.org/2009/02/17-rif-minutes.html#action01]

<trackbot> Created ACTION-707 - Add text to SWC indicating \"implementations should reject\" graphs with rdf:text etc [on Jos de Bruijn - due 2009-02-24].

<josb> we can change should to must

<josb> agree with Sandro

<sandro> If <pi>, 1 ≤ i ≤ n, corresponds to a DL profile and ui refers to an RDF graph that is not the RDF representation of an OWL (2) DL ontology, the document should be rejected.

<sandro> * If Profile does not have a single highest profile, the document must be rejected.

<josb> If Profile does not have a single highest profile, the document must be rejected.

csma: strong opinions re: should v must?

<josb> http://www.w3.org/2005/rules/wiki/SWC#Interpretation_of_Profiles

<sandro> ?PROPOSED: SWC will say imports MUST be rejected if the datatypes rdf:text or rif:iri are used in the imported graph.

jos: what about rdf:text used not as a typed literal?

<josb> If ui, 1 ≤ i ≤ n, refers to an RDF graph that uses a typed literal of the form "s"^^rif:iri or "s"^^rdf:text, the document must be rejected.

<ChrisW> agreed

sandro: torn between should and must
... "must" implies a test case

<ChrisW> ACTION: sandro add import rejection test category to test cases and add combination constant equivalence test case [recorded in http://www.w3.org/2009/02/17-rif-minutes.html#action02]

<trackbot> Created ACTION-708 - Add import rejection test category to test cases and add combination constant equivalence test case [on Sandro Hawke - due 2009-02-24].

<csma> PROPOSED: Change the test cases to "import rejection tests" and close issue-90.

<sandro> +1

<josb> +1

<csma> RESOLVED: Change the test cases to "import rejection tests" and close issue-90.

<AdrianP> +1

<csma> next item

<sandro> issue-57?

<trackbot> ISSUE-57 -- Does RIF specify an extensibility mechanism? -- OPEN

<trackbot> http://www.w3.org/2005/rules/wg/track/issues/57

<ChrisW> see: http://lists.w3.org/Archives/Public/public-rif-wg/2009Feb/0043.html

csma: motivation is to try to avoid "least common denominator" exchange
... e.g. PRD could interchange with CORE using some PRD syntax

sandro: does this allow the WG to add a new dialect in 2 years

csma: yes. PRD defines XSLT with fallbacks to CORE. Later, PRD++ defines fallbacks to PRD

sandro: but how does PRD get the new fallbacks?

csma: default fallback is reject. extensions have a fallback that is an XSLT

<AdrianP> the XSLT translation is a syntactic mapping. How does it solve the semantic differences?

sandro: can PRD++ fallback to PRD?

csma: may have to transform twice

sandro: advantage is simplicity, but cannot fallback across dialects (e.g. PRD - BLD)
... requires lots of fallbacks in instance doc

<ChrisW> can or can't be put in the schema?

sandro: thinks it works ok, but would like something more general

csma: doesn't handle "approximate" fallback with information on what is missing

sandro: how does it compare with xsi fallback?

csma: needs to review
... thinks no xslt possible with xsi fallback

<csma> <PRD++Elt fallback="PRD++->PRD">

<csma> <Do fallback="Do2And.xsl">

csma: PRD++ -> PRD -> CORE double fallback is possible if the first XSLT generates the second fallback

<AdrianP> this fallback solution might work from PRD to Core but not from Core to PRD

<AdrianP> so we can support round tripping

<AdrianP> this XSLT approach is a translator

chris: why put fallback in instance docs?

csma: because a RIF doc does not ID its dialect

csma: when you find an unknown element, try the fallback. Keep going until you know all the elements or run out of fallbacks (and then reject)

harold: 2 cases - "desugar" vs. semantics

csma: this proposal is for desugaring only
... can do desugaring on producer side without fallback, but need this proposal to do desugaring on consumer side

<AdrianP> for the fallbacks we would need to distinguish between "semantics preserving" fallbacks and fallbacks changing the semantics

<ChrisW> "default is reject"

sandro: will consumers be happy to do the fallback transforms?
... they really must do the transforms, or we don't have interop

harold: we should persue some fallback work, possibly in a next phase of work

<sandro> STRAWPOLL: Okay to require RIF consumers must implement fallback transformation

<sandro> +1 (to this or something like this)

<josb> 0

<gary> -0

<ChrisW> 0

<csma> +1

<AdrianP> 0 (not sure if we have enough time to do it now)

<Michael_Kifer> 0. I am not sure what the implications for the semantics are.

<Harold> +1 (although Should may be better than Must here)

<AdrianP> we would need to add a formal proof

<AdrianP> to the description of the XSLT translator

<csma> next item

sandro: have a general syntax unconstrained by dialect - common superset syntax spanning BLD, PRD, FLD

michael: promising direction, details need to be checked carefully

<AdrianP> details about the use of symbols need to be checked also inspite of future extensions of RIF

chris: eliminate stylistic debates

sandro: merged syntax has all the shortcut syntax

<Harold> Should we have less or more redundancy?

<Harold> "variables not needing the leading "?"

<Harold> sounds surprising.

<Michael_Kifer> unclear relationship to the shortcuts we introduced to augment the abstract presentation syntax

csma: is there a risk of this taking a long time to work out?

sandro: help with test cases, documentation, adoption, etc

<Harold> In a (syntactically) higher-order dialect we need to distinguish between likes(john,may) and ?p(john,may).

<Harold> (so likes and ?p need to be distinguished)

<ChrisW> PROPOSED: extend 5 mins

<ChrisW> RESOLVED: extend 5 mins

harold: too early to form an opinion
... can see several issues

next PS meeting is 11 EST on Fri on this IRC

<Michael_Kifer> I can't make it this fri.

<sandro> Harold, my thinking is that the ? is optional when the variable is declared (which might be always).

<ChrisW> ACTION: christian to put merged PS on agenda for Mar 3 [recorded in http://www.w3.org/2009/02/17-rif-minutes.html#action03]

<trackbot> Created ACTION-709 - Put merged PS on agenda for Mar 3 [on Christian de Sainte Marie - due 2009-02-24].

<ChrisW> AOB?