See also: IRC log
<hsolbrig> scribenic: hsolbrig
<hsolbrig> scribenick: hsolbrig
<hknublau> webex is not working for me right now.
<Dimitris> trying to connect to webex and getting errors
<Dimitris> holger I just made it, maybe you can try again
<Arnaud> hknublau, are you joining?
<Arnaud> https://www.w3.org/2014/data-shapes/wiki/F2F4
@arnaud: please check the wiki FTF list and make sure the expected attendance list is accurate
<hknublau> I dialed in via skype now.
<Arnaud> PROPOSED: Approve minutes of the 13 August Telecon: http://www.w3.org/2015/08/13-shapes-minutes.html
<ericP> +!
<ericP> +1
RESOLUTION: Approve minutes of the 13 August Telecon: http://www.w3.org/2015/08/13-shapes-minutes.html
arnaud: the FTF needs to start no later than 10 because lunch is fixed at 12:00 PM -- we can meet until 7:00 PM or so
arnaud: pfps raised question about what documents are to be published, hknublau suggests 3, but pfps only wants 1
... 3 documents -- primary, rdf turtle file and generated document
... what should we publish - one, two or all three?
kcoyle: do they have to be published together? They will all be available?
hknublau: there are normative references from the primary to the other documents
arnaud: it is possible to publish a document with references to working drafts, but would prefer all three. Can everyone review all three before publishing?
simonstey: I think we should publish the vocabulary with the main document, so all three should be published together.
pfps: adding the other two documents is going to be a lot more reviewing...
arnaud: the turtle and vocabulary are generated together. If you trust the transform it is only two documents
simonstey: what would it mean if we just published the draft?
... If we published the main draft and the vocabulary later, that would also work
<pfps> There is information in the turtle file that does not appear in the vocabulary reference document
arnaud: we will stick with the current schedule and then decide whether we feel comfortable publishing everything or just the first one
<pfps> committed to reviewing what?
arnaud: do we have two people who can commit to reviewing the spec?
pfps: there is information in the turtle that aren't in vocab reference
arnaud: lets focus on editors draft for the time being
... due date is in two weeks from now. Before FTF meeting
<simonstey> <-
<pfps> ACTION: pfps to review http://w3c.github.io/data-shapes/shacl-ref/ [recorded in http://www.w3.org/2015/08/20-shapes-minutes.html#action01]
<trackbot> Created ACTION-28 - Review http://w3c.github.io/data-shapes/shacl-ref/ [on Peter Patel-Schneider - due 2015-08-27].
pfps: I can commit
arnaud: I think arthur commited as well. We should have 2 or 3
arnaud: The test suite. Issues would better be illustrated with test and it could narrow down discussion. We need to have a couple of people working on the test suite...
<simonstey> afaik holger is actually doing this
<hknublau> These tests currently work: https://github.com/w3c/data-shapes/tree/gh-pages/data-shapes-test-suite/tests/features
<ericP> https://github.com/shexSpec/shexTest/ https://github.com/shexSpec/test-suite/ and docs at http://shexspec.github.io/test-suite/
ericp: The tests are written in ShEx, but Jose has a converter to SHACL
... We can convert them to SHACL.
arnaud: what do the tests in the repo do?
hknublau: manifest points to ttl file and checks output for T/F or specific results
... covers interesting features but are not exhaustive
arnaud: when we get to the FTF we should have someone ready to walk us through the suite, how to commit, etc.
<Arnaud> PROPOSED: Open ISSUE-79, ISSUE-80, ISSUE-81, ISSUE-82
<pfps> I'm fine with opening them all
eric: two implementations tested the proposed tests
<kcoyle> +1 open all
RESOLUTION: Open ISSUE-79, ISSUE-80, ISSUE-81, ISSUE-82
arnaud: Issue 79 came with a proposed resolution.
<simonstey> ISSUE-79
<trackbot> ISSUE-79 -- Cleaner separation between value checking and property iteration -- raised
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/79
<Arnaud> PROPOSED: Close ISSUE-79, per Holger's proposal
hknublau: Issue 79 allows logic to be reused and repurposed in different places
<pfps> so this is syntactic sugar for those few who write templates
<hknublau> +1
<simonstey> +1
<pfps> 0
<kcoyle> +1
+1
<Dimitris> +0.5
<ericP> +0
RESOLUTION: Close ISSUE-79, per Holger's proposal
arnaud: involves execution order. pfps says it shouldn't matter, kcoyle didn't get a response from dc-architecture list
kcoyle: without the "friendly front end" we probably won't get much response... ShEx?
<Arnaud> issue-76
<trackbot> issue-76 -- Specifying execution order and commutativity of AND and OR -- open
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/76
hknublau: two topics -- syntax and semantics. Syntax, we need to store in an RDF list because people expect order to be preserved...
... whether we want to do it semantically, the change should be small either way. If we change it, I would like to see a complete proposal
pfps: a lot of things aren't ordered, so surprised that holger wants this one to be
<Zakim> ericP, you wanted to ask if the shape level is an AND written as a repeated property while any nested ANDs are written as a list
hknublau: you would enter something on the screen vs. rdfs list
<pfps> by the same logic, entering constraints in one order should produce the same order, but that is not the case currently
pfps: ordering of AND's matter -- they produce different results
@arnaud: losing scribe
<kcoyle> scribenick: kcoyle
pfps: in shex there is a complex algebra for and/or, some domination so you can come up with the right answer if you vary the execution order
<Arnaud> STRAWPOLL: Close ISSUE-76, stating that execution order a) matters, b) does NOT matter
pfps: referring to shex member proposal
hknublau: question to Peter: execution order matters with AND
<simonstey> we are returning a set of error messages
pfps: if first produces error...
<pfps> users will care when things that can be fast take a very long tim
Arnaud: as a user, I don't care - if behavior between engines is different (e.g. reports one error not the other) - just matters that they produce an error
... even if sequence is different
pfps: users will care if what should be fast takes a long time
Arnaud: that's a quality of implementation question
pfps: depends on how trivial each branch is; common in queries; changing order influences execution time
simonstey: if execution order doesn't matter, it's up to the tool to optimize, rather than follow an order
<Arnaud> STRAWPOLL: Close ISSUE-76, stating that execution order a) matters, b) does NOT matter
<pfps> b
back to strawpoll
<simonstey> a) 0 b) 1
a} -.5 b) +1
<pfps> a) -1, b) +1
<Dimitris> a)0 b)0
<hknublau> a) 0 b) 0
<Arnaud> PROPOSED: Close ISSUE-76, stating that execution order does NOT matter
pfps: there is more to worry about if order does not matter; things are clearer if order is fixed
+1
<simonstey> +1
<pfps> +1
pfps: big issue is recursion through negation
<Dimitris> +0
<hknublau> 0
RESOLUTION: Close ISSUE-76, stating that execution order does NOT matter
<Arnaud> issue-65
<trackbot> issue-65 -- Consistency and cohesiveness of nomenclature (e.g., shapes, scopes, and constraints) -- open
<trackbot> http://www.w3.org/2014/data-shapes/track/issues/65
terminology cleanup; Peter sent many suggestions in email
Arnaud: what is a shape? what is a constraint? these need definitions
... options 1) stick to current draft and fix any problems in language or 2) move to Peter's draft definitions
... Holger replied: this could set us back some significant time
... but now is the best time to make terminology changes, before first public working draft
hknublau: terminology is consistent with shex and shapes
... don't see any benefit in changing
pfps: current nomenclature conflates shapes and constraints
... shapes that are used both as shapes and constraints
... shapes class gets used in two places for two purposes
... 1) as in shex, and 2) what shex calls a constraint
<pfps> it is actually very hard to figure out what is going on in the current editor's draft because the document is very hard to figure out
kcoyle: what would it look like if changed
<Arnaud> https://lists.w3.org/Archives/Public/public-data-shapes-wg/2015Aug/0096.html
simonstey: is starting an implementation; used to the terminology, and doesn't see it has a big problem
Dimitris: had some concerns in using shapes as classes, but haven't had time to read Peter's proposal
Arnaud: shape vs. class orthogonal; shape vs. constraint is more of a problem
hknublau: counter-proposal needs to include turtle file and examples
... if there is a problem, then need to see explicit examples
... but not worth the effort
Arnaud: paraphrasing Peter - better nomenclature would make editing of standard easier
hknublau: draft now has appendix with terminology; but these are editorial issues, not technical issues
... but changing names is a huge amount of work
pfps: Holger is apparently still working on document that is still under review.
Arnaud: document is stable; changes are being announced, only done after resolutions
... time to review = document is frozen; but announced changes are reasonable
pfps: have reviewed half, but not current version
hknublau: what has changed?: terminology consistency, editorial issues, nothing substantive
Arnaud: I did intend for the document to be frozen; however, if there are specific issues with documented changes, people can catch up.
... proceed; caution to Holger not to make broad changes to document; Peter continue review of version he has
pfps: working on 10 August version
Arnaud: re: nomenclature issue... Peter needs to produce an example to illustration change
<simonstey> +1
<Arnaud> trackbot, end meeting