RDF & SPARQL WG biweekly focused meeting

<pfps> where are the current tests?

<pfps> yes, but I can't find a pointer to the tests there

<ktk> ah like that. duno

make progress on test suites to move to CR 1

ktk: we want to see what is the goal in terms of tests

gkellogg: I marked up spec with references to normative statements
… there are tests in relation to semantics
… and others that I hope others could look at

pfps: there are syntax issues

pfps: the superman example has issues

<niklasl> w3c/rdf-tests

gkellogg: the repo for these things is rdf-tests
… where is the problem with the tests?

ora: we need to get these things fixed

gkellogg: the superman test does not work, problem with parentheses

niklasl: that test is rather negative

pfps: the test is incorrect because it uses rdfs++ entailment

pfps: rdfs+ is referenced in the manifest
… every task that has a triple term has to be checked
… I'll try to go through them
… would be nice if others can give a sanity check

niklasl: I'll try to do this

doerthe: I can also go over it
… after next week

ora: what else is missing related to test?
… to go to CR, we need complete test suite

pfps: Concepts does not need any test suite
… but semantics need

gkellogg: we should be linking to the test suite from the tests
… so we can easily find the tests related to normative statements
… the URIs are relative the spec suite location
… when looking at the rendered spec? it shows somewhere

pfps: apparently not in Firefox

gkellogg: maybe we did not merge the commit
… I'll take a look, there must be a PR with this

<pfps> what do look for in the source?

ktk: it would be good to have a nicely summarised list
… let's summarise: we have all the syntaxes (Turtle, Trig, N-Triples, N-Quads, RDF/XML) and semantics

pfps: the source of the spec mentions the tests

<gkellogg> https://www.w3.org/TR/rdf12-semantics/#ill-typed-tests

ora: but it is not rendered?

<gkellogg> https://w3c.github.io/rdf-semantics/spec/

gkellogg: it's possible it only appears in the editor's draft

<gkellogg> https://w3c.github.io/rdf-semantics/spec/#simple_entailment_properties

gkellogg: when you click on the word "tests" it shows the lists
… also, in the HTML version of the test manifest, it links back to this point in the spec

pfps: so we need to add the things as what's in Sec.5.3 in the editor's draft and that should work

gkellogg: all syntax tests have 1.1 versions of tests, semantics does not

<Zakim> niklasl, you wanted to ask about changing tests from opaque-to-transparent

ora: are there tests from 1.1 that must be deprecated?

<gkellogg> w3c/rdf-tests#190

<gb> Pull Request 190 Update RDF 1.2 semantics tests for triple terms not quoted triples. (by gkellogg)

niklasl: if I find time to deal with the semantics tests, should I make them transparent or what?

<AZ> s/what,/what?

gkellogg: please look at the RDF/XML tests and PR to the spec

What else is still needed for moving to CR? 2

pfps: just back up a second, if I approve Greg's PR, can we get someone to get it incorporated

gkellogg: yes sure

ora: now, we have a list of documents, what do we need to do to get to CR?

ktk: first batch and not first batch (see issue 167) is for prioritising

AndyS: we need to make sure it makes sense to do a batching

<Zakim> pfps, you wanted to discuss Semantics

pfps: AFAICT, semantics is closed, I'll ensure the tests are good, there are a couple of things that must be checked

ora: Pat Hayes sent questions and worries about the semantics of RDF 1.2
… there will be a call with Pat and the semantics editors
… other people could join but let us keep the group small

pfps: to summarise, Pat is concerned that triple terms do the same as what old reification vocabulary does, that is nothing
… why not do something better like N3

ora: we don't necessarily need to address all concerns, since Pat did not take part in last 2 years of discussions
… but we need to talk to him and take his words seriously

tl: I'd like to be part of the conversation with Pat H;
… there’s not much said about the reifier in the spec
… the domain of rdf:reifies is not defined
… the description of its semantics is not easy to find and it's a bit vague
… we may not be doing enough

gkellogg: there are 2 things about reifiers
… the concept that will be described in RDF Concepts
… there is little to say about it
… then there is the syntax with production rules
… also, there are tests that will need to be updated
… and the order of the list Andy made is not necessarily the publishing order of publications
… but in what order we need to take care of things

ora: is it possible that some people will not understand things we did with these new specs?
… we have to explain the new things well

tl: there is a mention of concretisation, it leaves me with more questions than answers

<tl> https://w3c.github.io/rdf-primer/spec/#section-triple-terms

tl: in Sec.3.5 in the Primer

pchampin: my position on reifiers is that they are deliberately under specified
… maybe we should be more explicit about that

<TallTed> s/s\/what,\/what?//

<niklasl> +1 to pchampin

pchampin: reifiers are a design pattern rather than something with a constrained formal meaning

<gkellogg> https://www.w3.org/TR/rdf12-concepts/#section-triple-terms-reification

gkellogg: Concepts makes it clear about reification
… in the Primer, we need to address the lack of explanation

tl: there are many things that are not normative about reifies
… there are non-normative statements about reifies in different documents and they don't help explaining it

gkellogg: the issue is about the semantics of reifiers and related things are because we have discussed over and over and could not get to an agreement
… this is something that should be addresseded after recommendation time

AndyS: we should move on and not return to the endless discussions

tl: I'll make an issue on Github and then the group can see what to do about it

ora: what else is missing?

gkellogg: first, we need to open PRs under three documents, and triage the issues specific to the doc

ora: can we do the triage next week?
… do the PR, then do the triage

<AndyS> Concepts has 4 noted open issues https://www.w3.org/TR/rdf12-concepts/#issue-summary

AndyS: looking at the list, there are issues related to the documents and we have a process to deal with what's related to the specs

Version announcement 3

gkellogg: we still need to decide whether we want to update the syntaxes to allow including a version
… in n-triples/n-quads, it's nice to have one line-one triple
… but the version stuff adds complexity

ora: adding a version announcement as a line in n-triples, n-quads conflicts with the principle of these formats
… there are lots of implementations that just split by line and assume there are only triples

pchampin: it's not the case, already in RDF 1.1 that one line corresponds to one triple all the time
… I appreciate what ora said, but I still prefer to have the version announcement

gkellogg: version announcement has to come before all triples
… we still have to fix what happens with the list of allowed versions and what happens if it's not valid

ktk: n-triples supports comments, really?

<gtw> easily missed because comments are treated as whitespace, not in the grammar.

gkellogg: yes but it's easier to leave them out as you can just ignore what starts with an #

AndyS: I am neutral on this topic
… there is value in being able to add the metadata into the file

<gb> Issue 169 definition of reifiers is non-normative and seems vague (by rat10) [needs discussion]

AndyS: n-triples allows concatenating files, which would not be possible with version announcement

gkellogg: we could mark it as a feature at risk
… so that people know that it may be left out

<niklasl> +1

<AndyS> +1

gkellogg: I can update the PR with that

<Souri> +1

ora: I am in favour of that

<AZ> +1

<james> +1

ora: any objection?

people: ....

ktk: should we decide who is going to talk to Pat

ora: is there an EXISTS call tomorrow?

AndyS: there is

<ktk> s/there are tested/there are tests/

<ktk> s/syntaxes to allow to