W3C

RDF-star

26 Feb 2021

Attendees

Present
pchampin, AndyS, thomas, ora, gkellogg, TallTed, olaf, gatemezing, james
Regrets
Chair
pchampin
Scribe
ora

Contents


<pchampin> scribe: ora

Review pending actions

<pchampin> https://github.com/w3c/rdf-star/issues?q=is%3Aopen+is%3Aissue+label%3Aaction

phampin: Test suites

andys: work in progress
... we can close the action, for now

<pchampin> https://github.com/w3c/rdf-star/pull/114

gkellogg: test suites can be difficult

SPARQL-star

pchampin: outstanding pull requests and issues

phampin: (not too controversial, mostly)

<pchampin> Subtopic: annotation syntax

<pchampin> https://github.com/w3c/rdf-star/pull/106

pchampin: annotation syntax

<olaf> https://pr-preview.s3.amazonaws.com/w3c/rdf-star/pull/106.html#sparql-star-grammar

olaf: propose not to extend the grammar
... put a note in the spec, call it a restriction
... counterexample (example 5)

<pchampin> https://pr-preview.s3.amazonaws.com/w3c/rdf-star/pull/106.html#example-5

olaf: cleaned up the grammar a bit, ready to be merged

gkellogg: there is a specific class of illegal examples, will add
... changes rendering subtly

<pchampin> PROPOSED: merge PR 106 after gkellogg adds the specific class on example 5

<AndyS> The syntax test suite has an "bad syntax" test for the path ":p/:q" case.

<gkellogg> +1

<AndyS> +1

<olaf> +1

<pchampin> +1

<gatemezing> +1

+1

<james> +1

<thomas> +1

<TallTed> +1

<gkellogg> the class name is “illegal-example”. I added a change note to the PR.

RESOLUTION: merge PR 106 after gkellogg adds the specific class on example 5

<pchampin> SubTopic: allowing for triples in the VALUES clause

<pchampin> https://github.com/w3c/rdf-star/issues/108

olaf: VALUES clause allows to bind values to variables; should extend to cover triples
... can create a pull request for this

<AndyS> Good idea.

pchampin: should we have a formal vote once the pull request is there?

gkellogg: there should be a vote for every pull request, but that can be a burden
... we can merge some time after the pull request is there, that gives folks time to comment

olaf: post on mailing list, leave three days to comment

<pchampin> PROPOSED: olaf to implement the proposal in issue 108 into a PR, and merge its after 3 days if no objection

<gkellogg> +1

+1

<olaf> +1

<pchampin> +1

<thomas> +1

<TallTed> +1

<AndyS> +1

<gatemezing> +1

<james> 1+

<mielvds> +0

RESOLUTION: olaf to implement the proposal in issue 108 into a PR, and merge its after 3 days if no objection

ghislain: are we allowing the use of the shortcut 'a' in the grammar? (scribe did not understand)

<olaf> https://github.com/w3c/rdf-star/issues/108#issuecomment-782853946

olaf: these are the parts of the grammar we propose to change

<olaf> https://pr-preview.s3.amazonaws.com/w3c/rdf-star/pull/106.html#expand-syntax-forms

gatemezing: this is fine

<pchampin> SubTopic: build-in functions to (de-)construct triples

<pchampin> https://github.com/w3c/rdf-star/issues/109

<pchampin> https://github.com/w3c/rdf-star/issues/27

pchampin: duplicate issues

olaf: built-in functions to allow us to access the components of a triple, and to test whether a bound value is a triple

(thanks gatemezing)

olaf: these should be like any standard built-ins

pchampin: should using a specific namespace be cleaner?

(pchampin hears voices in his head ;-)

gkellogg: this is an unofficial extension, should not co-opt an existing namespace; maybe we need a RDF*/SPARQL* namespace?

andys: these are like keywords

<AndyS> https://www.w3.org/ns/sparql#

pchampin: symmetry argument can be made, makes sense

james: symmetry argument is paramount
... where will you ultimately put these, that's where they should go now

olaf: upper case or lower case? No clear pattern in the SPARQL spec

andys: they are case-insensitive
... except for "a", "true", and "false"

olaf: propose to create a pull request, leave it open for a couple of days for folks to comment

<pchampin> PROPOSED: olaf to implement the proposal in issue 109 into a PR, and merge its after 3 days if no objection

<pchampin> +1

+1

<olaf> +1

<thomas> +1

<gatemezing> +1

<james> 1+

<AndyS> +1

<james> 1+

<mielvds> +1

<TallTed> +1

RESOLUTION: olaf to implement the proposal in issue 109 into a PR, and merge its after 3 days if no objection

<pchampin> SubTopic: removing the old version of the BIND operator

<olaf> BIND(<<:bob ?p ?o>> AS ?t)

<AndyS> FYI-- 1.1 spec Grammar note 1 is about case-sensitivity

olaf: meaning of this is that every triple that matches has to be bound to ?t
... this is for matching, not for construction
... should this be called something else, since the standard BIND can construct

<olaf> :bob ?p ?o . BIND(<<:bob ?p ?o>> AS ?t)

olaf: this would be a way to construct
... proposal is to remove this from the spec

pchampin: asserted triples only?

olaf: no, any triples
... in the draft, it is now only for asserted triples, unlike in the original version
... original draft assumed PG mode

gkellogg: this is much more complex than I thought

olaf: implementation complexity: agreed, it can be used in both directions (find or construct)

pchampin: the symmetry made it counter-intuitive

olaf: propose to remove it from the draft

<olaf> https://github.com/w3c/rdf-star/issues/6#issuecomment-783305045

gkellogg: it would be nice to capture related use cases

andys: consequence of this is that we do not need the bracket syntax (?)

<pchampin> bob ?p ?o . BIND(Triple(:bob, ?p, ?o) AS ?t)

<olaf> +1

andys: that would be preferable

<olaf> bob ?p ?o . BIND(Triple(:bob, ?p, LANG(?o)) AS ?t)

andys: bracket syntax may be convenient, but not necessary

<pchampin> BIND(<<:bob a :Person>> AS ?t)

pchampin: strange
... embedded triples are just terms

andys: this is an expression
... variables would be required to be bound
... could add it as a special case to BIND syntax, or it could be an expression that could be used anywhere

<AndyS> BIND(SUBJEC(<<:bob a :Person>>) AS ?t)

<AndyS> BIND(SUBJECT(<<:bob a :Person>>) AS ?t)

<AndyS> BIND(<<?s a :Person>> AS ?t)

<pchampin> BIND("a"^^?dt AS ?t)

andys: move to issues

<pchampin> PROPOSED: remove the old BIND (a.k.a. FIND) construct from SPARQL-star

<olaf> +1

<pchampin> +1

<james> +1

<gkellogg> +1

<gatemezing> +1

<AndyS> +1

0

<TallTed> +1

<thomas> 0

<mielvds> +0

RESOLUTION: remove the old BIND (a.k.a. FIND) construct from SPARQL-star

<AndyS> ACTION: AndyS Raise new issue for the BIND(<<:bob a :Person>> AS ?t) discussions

pchampin: the new BIND needs some more discussion

relationship between RDF-star and Notation3

thomas: Notation3 has formulae

<mielvds> +q

thomas: embedded triples would be similar?
... blank nodes?

pchampin: Notation3 scopes bnodes within the formula

<mielvds> I was just going to say

<mielvds> that i'' comment on this later, because I can't typ that fast

<mielvds> anyway, the people really involved are not on the call

<mielvds> N3 semantics hasn't been decided, but atm I would say thomas's interpretation is correct with the addition of the blank node thing

<mielvds> yes yes, I know, but the deep semantics people

<mielvds> the chairs I mean

<mielvds> but nesting in RDFstar is different from citing a graph

thomas: it is possible to have different IRIs for the same thing
... this is what formulae enable

pchampin: co-reference is not part of "core RDF"

thomas: "unique name assumption"

<mielvds> can we open a ticket on the relation w. N3? (Sorry for the mic thing)

pchampin: let's open a ticket

<pchampin> ACTION: pchampin to open an issue on he relation btw N3 and RDF-star

thomas: this is not the appropriate semantics for embedded triples
... cf. occurrences

<mielvds> I see N3 as a strict superset of RDF-star, so whatever semantics RDF-star would influence those in N3 (IMHO)

pchampin: I may disagree

<mielvds> bye

bye, stay safe

<gatemezing> bye!

<olaf> Thanks me if I have implemented them ;-)

<AndyS> bye all

<AndyS> rssagent, please leave

Summary of Action Items

[NEW] ACTION: AndyS Raise new issue for the BIND(<<:bob a :Person>> AS ?t) discussions
[NEW] ACTION: pchampin to open an issue on he relation btw N3 and RDF-star
 

Summary of Resolutions

  1. merge PR 106 after gkellogg adds the specific class on example 5
  2. olaf to implement the proposal in issue 108 into a PR, and merge its after 3 days if no objection
  3. olaf to implement the proposal in issue 109 into a PR, and merge its after 3 days if no objection
  4. remove the old BIND (a.k.a. FIND) construct from SPARQL-star
[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version (CVS log)
$Date$