Meeting minutes
Scribe?
andys: nothing on process and admin for today
andys:
andys: tf members should have been notified
zakimg, next item
Process and Admin
<AndyS> Issue list: https://
andys: today is about identifying issues
andys: I believe that the list above is complete (as of now)
andys: are these all clear
andys: this list hasn't changed in a while
andys: this mostly comes from an email from 2016
james: another issue is complexity - the substitution semantics is slow
andys: can you set up an issue for this, with examples
andys: performance issues can be tricky
james: here there are decided performance questions
james: I'll set up an issue if I know how
andys: I'll write something on how to set up an issue - after the meeting
andys: an issue should be somehow actionable
andys: the issue list is not closed
gtw: this is a list for discussion. we might not actually make changes to address them all
Proposed: the list at https://
Proposed: the items at https://
<pfps> +1
<gtw> +1
<AndyS> +1
<Tpt> +1
<gkellogg> +1
<olaf> +1
<james> +1
<tl> +1
RESOLUTION: the items at https://
andys: new items need a good description
pfps: the TF will look at anything with an adequate description
andys: the WG needs to be careful about making changes that don't have adequate support
Confirm identified issues
How to address errata
james: we have said that we want to gather results, but what are we going to do with them?
james: there is a repository and
james: and a script that can be run to get the results
james: I think that either a table or page per implementation showing the results would be good
pfps: I'll try to run the script on my machine for Virtuoso, QLever, Blazegraph, and MillenniumDB
james: I didn't run into any errors, even syntax errors
pfps: My recollection is that implementations don't produce syntax errors, but instead do something other than the standard
andys: the tests should all have answers
james: in my testing I didn't check the results
andys: criteria for #130
<gb> Issue 130 vocabulary to refer to the individual nodes in a reified triple term (by rat10) [needs discussion]
<AndyS> w3c/
<gb> Issue 130 Criteria for evaluating errata changes for EXISTS (by afs) [spec:bug]
<gb> Issue 130 vocabulary to refer to the individual nodes in a reified triple term (by rat10) [needs discussion]
pfps: [describes several
pfps: [described several criteria]
andys: it's not actually textual substitution
pfps: agreed, I fixed the issue comment
james: are we discussing these as preferences or requirements
pfps: in my view these are preferences
andys: preferences as well
andys: please add any other desirable criteria to the issue
james: why should these not be requirements
pfps: because they conflict
pfps: a way of proceeding is to select some and see if they conflict
james: we will be making choices
<james> https://
Topic for next time
james: I believe that this is pertinent to the discussion and doesn't have complexity issues
pfps: right now SPARQL queries outside of EXISTS are not correlated
james: correct
<AndyS> "Unnesting Arbitrary Queries" -- Thomas Neumann and Alfons Kemper
andys: SHACL uses parameterized queries which are related
andys: SHACL has use cases that test for EXISTS (possibly negated) on single triples
james: can
james: can test cases be developed from SHACL
andys: yes, but it may be work to make test cases than can be easily understood
andys: next time?
andys: go though test cases and determine coverage?
<gtw> where are the most up-to-date test cases?
andys: we could also look at the PR
tpt: the test suite should cover all the issues
andys: so next time review the tests
<AndyS> afs/
tpt: can we name the tests to include which issue they address?
andys: sounds good
<AndyS> +1
<james> i need to leave for today. goodby until next time.
andys: next meeting next week