See also: IRC log
<scribe> scribenick: kcoyle
<Arnaud> PROPOSED: Approve minutes of the 14 Sept 2016 Telecon: http://www.w3.org/2016/09/14-shapes-minutes.html
Bart pinged - ok with Issue 137 resolution
RESOLUTION: Approve minutes of the 14 Sept 2016 Telecon: http://www.w3.org/2016/09/14-shapes-minutes.html
Tied for Tuesday and Wednesday, with some not being able to make one or the other
Can do one or the other, or alternate weeks, one Tuesday, one Wednesday
TallTed: every other is not a good idea (Eric says the same thing)
<AndyS> Help editors to attend?
Arnaud: when will we meet next week?
<TallTed> +1 Sept 27, encouraging early agenda dispatch
<Arnaud> PROPOSED: next meeting on Tuesday 27
RESOLUTION: next meeting on Tuesday 27
<trackbot> issue-71 -- SHACL Endpoint Protocol -- open
Arnaud: reached out to JimA regarding OSLC; ok with closing it
<Arnaud> PROPOSED: Close ISSUE-71, as is - not enough time left to tackle this issue in this version of SHACL
RESOLUTION: Close ISSUE-71, as is - not enough time left to tackle this issue in this version of SHACL
<trackbot> issue-105 -- SHACL SPARQL constraints depend on namespaces in a graph, which is not defined -- open
<Arnaud> PROPOSED: Close ISSUE-105 adopting the so called "decomposed prefixes in Turtle" proposal drafted by Holger in https://rawgit.com/w3c/data-shapes/bf5efb571196cf3c6b63137302c9c13a369e3466/shacl/index.html#sparql-prefixes
RESOLUTION: Close ISSUE-105 adopting the so called "decomposed prefixes in Turtle" proposal drafted by Holger in https://rawgit.com/w3c/data-shapes/bf5efb571196cf3c6b63137302c9c13a369e3466/shacl/index.html#sparql-prefixes
<trackbot> issue-68 -- pre-binding not defined in SHACL spec -- open
AndyS: discussion on-going; key is to go for simplicity; blank nodes still a pattern because act as variable
... bnodes are treated as constants
... no entailment happening
Arnaud: good that discussion as resumed; not sure though about time frame (winter holidays coming up)
... need to be at CR by early 2017
... what are the alternatives? what if we cannot agree?
AndyS: direction I outlined are the ones that will work best; uses of exists/not exists in actual use not that complicated
... group cannot make decision, not a working group
Arnaud: process that exists makes it hard to do revision of specs (that may have errors)
... put in our spec "sparql broken"?
AndyS: set a good example - community can decide to revise spec
<Zakim> ericP, you wanted to say we can add an errata as well
AndyS: two problems: 1) bnodes (usually treated as constants) 2) things that dont make sense to write in the syntax
ericP: Arnaud proposing add note to our spec; but then people have no pointer from sparql to our spec
... could add to sparql errata
AndyS: community group will come up with a report, can point to that
... process: write to sparql comments list, and gets added to errata
... errata doc just collects notes for next group
Arnaud: sounds good/legal. How do we highlight that in SHACL?
AndyS: point to community group document
Arnaud: would be a normative reference
AndyS: encourages everyone to join community group discussion
Arnaud: w3c getting serious about test suites; i18n needs to be checked; also security
... only 50% of working groups even have a PLAN for test suite;
... needed for CR
<ericP> ericP: the ShEx test suite is pretty methodical. has reasonable coverage testing.
<ericP> ShEx test suite
ericP: shex test suite had translations to shacl, but shacl has changed; there are about 700 shex/shacl tests
... doing a coverage analysis, e.g. or inside an and; node constraints in various positions
... depends on jose to do translation
Arnaud: encouraging that there is a large body of tests; doesn't have to be complete to go to CR, but plan and framework has to be there
<ericP> ShEx manifest
<Arnaud> trackbot, draft minutes
<trackbot> Sorry, Arnaud, I don't understand 'trackbot, draft minutes'. Please refer to <http://www.w3.org/2005/06/tracker/irc> for help.
ericP: quick tour - data types, node kinds, etc.
... (geeks out about floating and other functions)
Arnaud: have to close all issues
<trackbot> issue-177 -- Abstract Syntax is disconnected from concrete syntax -- open
ericP: need to catch up
hknublau: language tag issue is resolved
<ericP> kcoyle: we've done extensive edits based on recieved comments. please read editor's draft.
draft has been updated based on comments
don't yet have error levels
ericP: error levels don't have impact on semantics
... in shex level is just annotation property
... so we can add to AS and note that it doesn't have impact on semantics
ericP: how to test error functionality?
hknublau: each constraint violation has point to constraint component - so message will be different in different implementations, don't need to test
ericP: don't know what error message will be returned - can't know what the error message will be
hknublau: but those are different errors, so they would all need to be returned
... the results ontology returns one object, link to focus node
hknublau: says what has to be returned; footnote says some implementations return t/f
... each condition has a returned validation message
ericP: in shex this is just informative; doesn't say what failure is
... so need to add to test suite
hknublau: already has this in test framework, saved returned results as part of tests
<hknublau> Can't hear anything anymore.
<hknublau> Meeting still ongoing?
<hknublau> WebEx doesnt seem to connect me any more.
<hknublau> I try reconnecting, but it hangs now.
ericP: we will need to add test messages to AS
Arnaud: would like to publish more often; so need to keep two drafts in sync
... need to show "wide review"
ericP: need to show what communities you reached out to
Arnaud: INRIA has an implementation of SHACL
... possibly just core; will find out more
<trackbot> issue-178 -- Should sh:message be permitted at constraints, too? -- raised
hknublau: no way to customize message on core constraints
... propose to generalize sh:message - to be usable on any constraint
<Arnaud> PROPOSED: Open ISSUE-178
<Zakim> ericP, you wanted to say that having defined, structured error responses is motivated by this use case
ericP: one main reason to have defined structured error messages; is to have structured responses
... the hook already exists;
hknublau: this gives the power to the author of the shape; doesn't depend on engine
ericP: this could be an application-specific feature; but this makes it a standard feature
RESOLUTION: Open ISSUE-178
ericP: my concern is - what happens if there are multiple constraints?
hknublau: this isn't always useful when there are multiple constraints - so author of shape can choose to structure shape differently
ericP: can work around it by restructuring constraints around error messages
... what to do when every node constraint gives the same message?
hknublau: allow this; let author create usable message
Arnaud: can't prevent people making mistakes
ericP: document this well
hknublau: peter sent email ~ month ago; we need to respond to public mailing list comments
Arnaud: best to have "informal-ish" process; just be careful to represent accurate WG position
hknublau: will send email linking to page with responses
Arnaud: brings up virtual f2f again - think about it as a way to finalize spec
<Arnaud> trackbot, end meeting