See also: IRC log
<kcoyle> webex is twirling - is the phone line working?
<scribe> scribenick: dimitris
<Arnaud> PROPOSED: Approve minutes of the 8 October Telecon: http://www.w3.org/2015/10/08-shapes-minutes.html
arnaud: propose to approve minutes of last week
RESOLUTION: Approve minutes of the 8 October Telecon: http://www.w3.org/2015/10/08-shapes-minutes.html
<Arnaud> PROPOSED: Open ISSUE-102
<ericP> +1
<Arnaud> issue-102
<trackbot> issue-102 -- Some defaults cannot be explicitly coded -- raised
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/102
<kcoyle> +1
+1
<TallTed> +1
<Labra> +1
arnaud: this is a reasonable question to ask
aryman: I don't like the title, we shouldn't imagine that every template should have a default value
... most of the constraint are similar
... default only makes sense if something is required
tallted: by declaring this cardinality we say that by default it is 0 - unbound
aryman: we also do not have default constraints
arnaud: how should we rename this issue?
aryman: we also have min length, max length
kcoyle: happy to change the title a bit
... i would like to say my max count is unbounded
<Arnaud> how about: missing a way to explicitly state unbound cardinality and open shapes
<aryman> +1 as amended
RESOLUTION: Open ISSUE-102, changing the title to: missing a way to explicitly state unbound cardinality and open shapes
arnaud: Eric can you give an update on the test suite?
<ericP> http://shexspec.github.io/test-suite/
<ericP> https://github.com/shexSpec/test-suite
ericp: Jose gathered a test suite for ShEx some time ago. there are two test suites and I checked which one is applicable to SHACL
... most of the tests are but I have a question
<ericP> <myShape> { :p1 xsd:integer, :p2 xsd:float }
ericp: what is the representation of ShEx in SHACL and I was trying to figure out a simple conjuction
... and not sure how to translate in SHACL
aryman: they mean the same. Should we say that a given property can appear only once?
<ericP> <myShape> { :p1 xsd:integer | :p2 xsd:float }
ericP: do I represent them as a constraint with an AND?
... do I split that in two shapes?
aryman: I think yes. This is related to issue-95
Arnaud: so the status is you investigated how you can re-use the existing tests and you are looking how to translate automatically
<aryman> WebEx just dropped my line.
ericp: once these things get resolved it will be easier to translate them in SHACL and have variations
<aryman> Back in now.
arnaud: the use of and does not need to be explicit
... karen, are you also importing tests?
kcoyle: we don't have people that know shacl and no way to test our shacl
arnaud: how many people are planning to implement shacl
... TopQuadrant is doing it, Open Link Software will but not sure when, anyone else?
dimitris: I am also working on an implementation
aryman: I know Clark&Parsia might be interested
arnaud: Peter isn't here so this will have to wait
<trackbot> issue-95 -- Proposed simplification and clean up of template mechanism -- open
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/95
<aryman> why use sh:scopeClass for injection?
aryman: holger uses the term scopeClass to inject contraints, is this overloaded?
hknublau: sh:scopeClass is consistent to the validation tools
... with the same logic they can validate a constraint or an instance
aryman: you are arguing on code-reuse but scopeClass has a very different meaning
... and these are very different use cases
... the re-use of code shouldn't affect the language design
hknublau: it is consistent and affects a very small number of people
aryman: can you provide the meaning of sh:scopeClass in the spec?
arnaud: Arthur's request is reasonable, can you please provide it, either as link to the spec or on mail
... let's leave it for next week
<Arnaud> issue-86
<trackbot> issue-86 -- Associating shapes with ontologies or vocabularies -- open
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/86
arnaud: Dimitris made a proposal to resolve but Peter objected.
... we should recognise that people have different use cases
hknublau: looked into this yesterday and even the design with sh:shapesGraph is difficult with users
... dimitris suggests to add another keyword for another scenario. I am not totally against this but it get's compicated
... class definitions need to exist in the shapes graph and this is already too much to explain
... you can use owl:imports and a tool can identify it is a shapes graph
arnaud: a diagram can show this operations.
... dimitris make a more simplified proposal
<Arnaud> issue-98
<trackbot> issue-98 -- Can property constraints also apply to focus nodes? -- open
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/98
hknublau: my proposal is to allow constraints on focus nodes as well
... (summarizing the proposed changes)
arnaud: Peter proposed that it would be simplified further if we look at ShEx
... but this is not a stopper for now
aryman: I like where this is going. regarding syntaxt we shoudn't declare sh:constraint
hknublau: we examine that later, not sure about it now
aryman: this is similar to ResourceShapes. We should stop considering code but make language simpler
hknublau: code helps me identify consequences
arnaud: that's fair, we can agree to this and then Arthur can raise different issues
<Arnaud> PROPOSED: Close ISSUE-98, adopting Holger's proposed changes
<aryman> +1
<hknublau> +1
<aryman> +1
<kcoyle> +0
<ericP> +0
+1
<Labra> 0
<TallTed> +1
<Arnaud> - Rename sh:allowedValues to sh:memberOf (or sh:oneOf)
hknublau: we also need to decide how we rename sh:allowedValues
<aryman> Propose sh:in
aryman: sh:in can be used
<hknublau> ok for me
dimitris: what will in check in the case of focus nodes?
<TallTed> +0
<kcoyle> "in" sounds good
hknublau: focus nodes can be literals as well so it can be any value
+1 to sh:in
RESOLUTION: Close ISSUE-98, adopting Holger's proposed changes, renaming sh:allowedValues to sh:in
aryman: one of the oslc use cases is to use http get to a resource. that triple has to be in the data graph
... we need a triple in the data graph and we don't care about the direction
<aryman> the OSLC use case requires that the triple linking the resource to the shape be in the data graph
hknublau: we can call that triple pollution, this triple should go in the shapes graph, it has nothing to do with the data. it also consistent with the design of scopeClass
tallted: This makes the shape less re-usable
... it seems odd to have it in the shape definition
arnaud: people have different use cases
kcoyle: there is a number of different meanings
<TallTed> +1 aryman's comments
aryman: the shape is a type definition for data. how can the shapes graph contain links to the resources since resources can be created / deleted dynamically?
... the triple has to be present in the data graph
hknublau: I raised it after peter's proposal
... shapes can be in separate graphs and combined with owl:iimports
<TallTed> so this is an inverse-relation for sh:nodeShape -- but does not *belong* in the shape itself (though there *might* be a use case of which we're unaware, so we should perhaps not prohibit that)
<TallTed> I'm OK with having sh:shapeNode inverse of sh:nodeShape -- that's fine -- sometimes expression can only be in one direction
<TallTed> this statement can be made in any graph -- using sh:shapeNode does *not* mandate that the triple be in the shape (or shape graph)
arnaud: let's stop for now and wait for Peter who raised this issue
aryman: the notion of reversing the relation doesn't make sense
tallted: some tools have direction limitations
aryman: this adds complexity to get data and need to always create union sparql queries
tallted: you can build a reasoner to handle the reverse. there are lot's of ways to handle this
<hknublau> I generally dislike inverse properties.
<ericP> +1 to +1
<ericP> sorry, +1 to Aryman's point
arymam: rdf is a way of encoding statements, why do we need to use inverse predicates for everything
<Arnaud> trackbot, end meeting