Meeting minutes
#390 as replacement for #362 (ex namespace)
elianaP: There is a PR to switch namespaces for examples,
… this ended up hitting a line-endings issue making a very large diff
… Nick had made a separate PR replacing the original.
<simonstey> w3c/
HolgerK: It seems the new PR touches the old test cases, but we should probably not touch those.
ajnelson-nist: There is a way to universally change line endings with .gitattributes, but I (and elianaP ) can't recall if there's a filepath constraint to change this.
<simonstey> w3c/
elianaP: Suggest to change close 390 without merging,
… and work with 362 instead.
AndyS: We may want to switch tests into 1.0 and 1.2 tests.
HolgerK: They are already split into separate folders.
HolgerK: Suggest someone re-apply Nick's changes into new branch.
<AndyS> data-shapes-test-suite/tests/ is not obviously 1.0
elianaP: will apply the changes.
HolgerK: Note there is also a document for the test suite from 1.0. Was NOT copied into 1.2 space.
… I didn't know what to do with this document.
… We don't have an editor of that document.
… We could have the document get brought over to 1.2 to describe the 1.0 vs. 1.2 tests.
AndyS: It's not clear if tests are changed from 1.0 to 1.2. They do have the same test name.
AndyS: When I implemented SHACL, I had to look at the test cases, because they had more information.
elianaP: Do we need a change log comparing the two folders?
… Would that help implementers?
simonstey: It would help people find the test cases that are new.
ajnelson-nist: Testing 1.2 implementations against 1.0 test cases - 1.2 implementations should continue to pass 1.0, yes?
(group noted yes)
<HolgerK> LOL
bergos: It feels straightforward to run tests against 1.0 and 1.2 folders.
<simonstey> new
ajnelson-nist: Noted some tests have been changed since copying into 1.2 test suite;
<simonstey> old
simonstey: Are we permitted to change the folder name for 1.0 test cases?
elianaP: Unsure if process permits.
caribou: We could configure the URI to be generated from a different directory.
<caribou> SHACL 1.0 test suite and implementation report
elianaP: Instead of renaming folders, is it helpful to add a README in each test folder?
simonstey: The folders already have these READMEs.
elianaP: If we are not renaming folders, and we have instructions ... is there an issue remaining?
AndyS: You somewhat lose the ability to see the new tests.
bergos: Since READMEs are first things you see when you go to the repository, they should provide enough navigation hints.
TallTed: Anything outside of the /TR directory can be changed.
elianaP: Suggest to rename the folder.
caribou: Can update the redirect on the GitHub repository.
Allowing lists
elianaP: There were a few more topics in conversation of 407.
bergos: New idea;
… for those not happy with supporting lists, we have node expressions.
<TallTed> w3c/
add support for sh:TripleTerm to sh:nodeKind and allow lists
(ajnelson-nist couldn't capture all of the new idea)
HolgerK: Not seeing the need for different OR combinations with triple terms combined with others.
… would prefer to not complicate it further.
… Not seeing enough use cases for this.
simonstey: Justifiable to not include this, as Holger said;
… there seem to be some cases that would be AND-ed instead of OR-ed.
… TripleTerms seem different enough to justify not including them in the OR-grouping.
elianaP: Let's take 1 or 2 weeks for this vote.
simonstey: The list form is sleeker and easier reviewability, BUT would present backwards-compatibility issues.
elianaP: asked Thomas to include new idea in comment.
elianaP: Taking 2 weeks for this vote, to account for vacations.
HolgerK: The old working group would take a straw poll at this point, but we seem decently close either way in this case.
… we haven't done a straw poll yet.
elianaP: Let's not do a straw poll now, but it's an option we can use in the future.
Meeting conclusion
elianaP: Reminder, please look at Issue about triple terms and lists.
AndyS: Started work on formal model for rules and what evaluation means.