Meeting minutes
General discussion on timeline
nicholascar: what to do on remaining 24 core tickets + 4 sparql
Holger: 3 tickets on agenda are most important, others mainly editorial; can live without. Sparql mostly editorial, some prebinding q's to discuss with Andy
… potentially start on node expressions in Sept
<TallTed> core
<TallTed> SPARQL
nicholascar: focus on core; there has been interest in other work e.g. shacl UI but leave for now
HolgerK: node expressions is a next logical item to pick up; happy to become an editor and push this work to try finish by year end
HolgerK: shacl compact syntax are also waiting on core
nicholascar: what is the expected process on establishing community groups; while WG is in session?
HolgerK: potentially add label for "postponed" tickets in core and reevaluate later
bergos: on postponed tickets, there's one ticket I'd like to pick up number 192 i.e. not postpone
<elianaP> w3c/
HolgerK: others can pick up issues/make PRs, not only editors
nicholascar: propose tickets without an assignee soon be changed to "postpone" status
PRs available for major SHACL Core features
PR 395 on issue 160 Allow lists at sh:class and sh:datatype
PR 399 on issue 172 -- support for sh:closed sh:ByTypes
HolgerK: problem with sh:closed, only can directly close props directly at shape, did not consider inheritance, walk hierarchies etc.
… PR adds third value for sh:closed, true/false/sh:ByTypes
… if value is ByTypes, system must walk hierarchy and look at classes in hierarchy
nicholascar: why named ByTypes?
HolgerK: because looks at types (rdf:type) first, then looks at subclass of, sh:node, and then target class triples
nicholascar: potentially there is a better name, to think about
bergos: more discusson on the current name in the ticket
issue 173 reification for sh:deactivated, sh:messsage and sh:severity
<TallTed> s/topic: PR 160/subtopic: PR 160/
HolgerK: can currently only declare message on single shape
<TallTed> s/Topic: PR 173/subtopic: PR 173/
HolgerK: same issue with deactivated. Reification now helps with this; for this triple specify activated etc.
ajnelson-nist: does reification help with deactivate property shape but it's solely named as a blank node
HolgerK: recommendation to always use IRIs for property shapes; and that people should add rdf:type for property shapes
RDF 1.2 support
bergos: older PR for adding support for RDF 1.2 triple terms PR 368
… all concerns addressed; what is status? HolgerK has approved
<TallTed> s|s/Topic: PR|s/topic: PR|g
nicholascar: one other reviewer on call, TallTed , any thoughts on PR
TallTed: haven't seen PR recently, will review in next day
nicholascar: is PR in agenda built off 368?
bergos: keen to hear who is interested, should we discuss these now?
nicholascar: any other agenda/other items to discuss prior?
… None; continue with RDF 1.2 discussion
bergos: use case of targeting within triple terms
… puts focus node on the triple term itself,
… can use sh:inversePath rdf:reifies to go into the shape
… other targeting concept: is it a reifier; as could have different targets.
… two different things that would need be added for the targets, first filter then validation.
… triple part out means. Focus node is triple term, and want to go out
… other way round is matching in to triple term. Triple part in, specify which parts of the triple term to match, then use same properties for target matching,
nicholascar: to summarise, is it, target triple term, then triples part in or out?
… are there systems / use cases with cascading sets of reification?
ajnelson-nist: can think of examples e.g. disputes "he said x she said y" going a few levels deep
TallTed: will not know real use until people start using more
… urge caution until we know how people use it more
nicholascar: are nested/cascading cases going to be complicated, can we make it any better
ajnelson-nist: can look at inverse path with zero or more as an example of difficult cases
… question on traverse out of triple term. Talks about ex:age, and a version. Where is the age in the example?
bergos: age is the focus node at the end, match triple term that is the subject Bob, match triple term, traverse out with predicate
bergos: David made proposal to create separate predicates; I wanted to avoid this, matching on triple term is a little "hidden" but better to keep path as main tool for traversing
nicholascar: so can use the triple part in / out ?
bergos: yes you could do this with a sequence path, not covered in current examples.
nicholascar: is it necessary to use sh:subject/sh:predicate/sh:object
<ajnelson-nist> https://
ajnelson-nist: would clash with RDF 1.2; they use ttSubject etc.
mgberg: they only use this for classicising
elianaP: to clarify example; version is tied to the predicate, not the triple term?
bergos: versioning ontologies as an example, can version predicates
elianaP: clarifying for example, so e.g. age works the same way in version 1 & 2 so can be used
nicholascar: q to mgberg ttSubject etc. can not be used, wrong purpose?
mgberg: would need to check, not clear intention of spec
bergos: we need the predicate and the class as a value e.g. sh:predicate sh:Predicate; can cause issues if mixing namespaces e.g. rdf & rdfs
<mgberg> https://
nicholascar: if the purpose of inventing the predicates in the 1.2 namespace was not for this, perhaps other spec authors should reconsider or tell us why it might be wrong
mgberg: idea is to take something in 1.2 and put it into 1.0 graph; notes on round tripping, there are caveats
… conceptually they are close
nicholascar: as thought experiment, how to query for this in sparql if not using SPARQL 1.2 syntax
nicholascar: how many new things: target triple term, is reifier, in & out (+ classes)?
bergos: isReifier is not absolutely necessary, others yes all required for traversal
nicholascar: is there a single concept that would combine in/out
bergos: yes could use single property with qualifier, but think better not to
<TallTed> we're over time.... we should strive to end 5 minutes before the hour, as all w3c calls are encouraged to do
bergos: quite used to the in out style, not sure if there's an easier way to define it. No strong opinion on naming but think the basic structure is good
elianaP: lets try to include less esp. in relation to triple terms, go for simplicity
mgberg: core PRs or issues, go ahead and make PR, is this procedure?
nicholascar: yes claim core and sparql PRs