See also: IRC log
<ipolikoff> scribenick: pano
sub-topic: Disposal of raised issues
<ipolikoff> PROPOSED: OPEN issues issue-232, issue-233
RESOLUTION: OPEN issues issue-232, issue-233
ipolikoff: Sandro, could you introduce yourself
sandro: [introduces himself]
... possibly replace Eric from W3C side.
UNKNOWN_SPEAKER: There is a privacy and security questionnaire
... I think your answers to this questionnaire can lead to a privacy and security section
ipolikoff: How do we answers those questions, logistically?
sandro: usually somebody from the groups takes this on and answers these questions and a section is created based on that
TallTed: do you know of an example which we can look at?
ipolikoff: Essentially an objection was raised that a validation report is described in a procedural manner
hknublau: I've drafted a response.
... I believe this response clarifies, if not completely resolves this issue.
... I believe the mathematical style vs our style (basically for implementers) is a matter of taste
... it's always a trade-off.
TallTed: this section is going to have two main audiences. The implementers and the people who are trying to understand what the validation engine has done. I thought we stepped away from describing the validation and moved towards "conforming to the shape or not" ?
ipolikoff: yes, but we're still describing the validation report
TallTed: If we're going to blend validation and conformance I could live with that, but my point is more about the audience of this topic. The people who are going to be reading these reports need to be able to understand what they get back.
... we could add an example of a simple result.
hknublau: there are lot's of example snippets
TallTed: We should provide a complete example, definitely an example that covers all the MUSTs .
hknublau: If you look at section 1.4 at the end it shows the complete validation report
TallTed: it may just be a question of highlighting the example some more
... what do the different color boxes in the spec signify?
... it would be helpful to add a color box specific to the validation reports, so they pop out
<ipolikoff> PROPOSAL: CLOSE ISSUE-225 as addressed in section 4 - clarifying the validators always produce/new/result nodes, no change in the style of writing, add color to the validation results to make them stand out
<sandro> +0 sounds reasonable, but not familiar enough with issue
RESOLUTION: CLOSE ISSUE-225 as addressed in section 4 - clarifying the validators always produce/new/result nodes, no change in the style of writing, add color to the validation results to make them stand out
<ipolikoff> PROPOSED: Approve minutes of the 15 Feb 2017 Telecon: https://www.w3.org/2017/02/15-shapes-minutes
RESOLUTION: Approve minutes of the 15 Feb 2017 Telecon: https://www.w3.org/2017/02/15-shapes-minutes
hknublau: I've created a draft response which strongly overlaps with the issue-225 response
TallTed: on the wording of the response, I think exclude is the wrong word.
sandro: disallows is a possibility
<ipolikoff> PROPOSAL: CLOSE ISSUE-228 by ensuring cardinality 1 is specified for all relevant properties
RESOLUTION: CLOSE ISSUE-228 by ensuring cardinality 1 is specified for all relevant properties
hknublau: I also drafted a response to this issue.
... so, just a minor edit from this issue, and this email response
<ipolikoff> PROPOSAL: CLOSE ISSUE-229 as described in https://lists.w3.org/Archives/Public/public-data-shapes-wg/2017Feb/0037.html
RESOLUTION: CLOSE ISSUE-229 as described in https://lists.w3.org/Archives/Public/public-data-shapes-wg/2017Feb/0037.html
hknublau: a couple of people on the public list wonder in what way the case in which the shapes graph itself is not well-formed can be handled
... [describes proposal in https://www.w3.org/2014/data-shapes/track/issues/233 ]
ipolikoff: there was an additional email by Peter that suggests there must be a feature that validates the shapes graph itself
hknublau: that's right. I am not keen on that. It increases the workload and we don't have the means for that currently.
... we would basically raise the bar for any implementation to pass the conformance test.
ipolikoff: there's two issues. extra work for the WG and extra work for the implementers
TallTed: we're basically suggesting a linter for SHACL
... we have made requirement, so this is problematic.
hknublau: But, we haven't made such a requirement.
TallTed: ah ok, I misunderstood, we should then make it clearer that this is not a must
... The engine should report what it has found for testing the shapes graph.
hknublau: yes this could just be a boolean flag
sandro: it would be nice to have information on why a shapes graph invalid if it is invalid
TallTed: this check should be fairly uniform, you would want to have the same result for different engine implementations
hknublau: I'm sure this feature will emerge as an open source project.
sandro: do we have a list of postponed features? This could be a place to put features like this.
<ipolikoff> PROPOSAL: CLOSE ISSUE-230 by introducing sh:shapesGraphWellFormed flag which is set to true in the sh:ValidationReport object if the engine is certain that the shapes graph is passing all syntax requirements and adding a postponed feature for validation of the shapes graph
TallTed: in RDF world, having a null value is problematic. Setting one explicitly leads to all kinds of unexpected results, so I wouldn't want to go with that as a third option.
sandro: isn't it ok to ommit it then?
TallTed: leaving it empty then it is unclear if it is unknown, or undetermined
... indeterminate is "I ran into something that I don't know how to check"
hknublau: I don't agree that the open world assumption is a problem here.
TallTed: This *is* open world. An unset column cannot be seen as determinate
dallemang: then sandro's comment becomes relevant again: if you don't get true on the conformance test then it's clear that you should proceed with caution
hknublau: So should I put in the spec that if the shapes graph is invalid the value is false or not?
dallemang: it's up to the implementation to handle what happens after validation leads to a false value
RESOLUTION: CLOSE ISSUE-233 by introducing sh:shapesGraphWellFormed flag which is set to true in the sh:ValidationReport object if the engine is certain that the shapes graph is passing all syntax requirements and adding a postponed feature for validation of the shapes graph
hknublau: there was a related comment by Peter on recursion that relates to this slightly. Handling of this is currently unspecified.
... we could use a similar flag strategy for this
TallTed: to describe it in text is possible, but once you want to describe this as expressing engine capabilities in a specific way, these should be part of the spec
<hknublau> Great, could you add "comment" for rdfs:comment ?
yes I will!
hknublau: it seems as if Andy is making progress on this issue. I hope this is not a blocker for CR.
dallemang: What work do you see that needs to be done on this?
hknublau: it seems that there haven't been any significant problems found yet
<sandro> sandro: be sure to indicate if comment is blocking
<TallTed> trackbot, end meeting