RDF Data Shapes Working Group Teleconference

29 Mar 2017

See also: IRC log


sandro, simonstey, hknublau, ipolikof, TallTed, Dimitris, dallemang


<TallTed> scribenick: ipolikof


<TallTed> PROPOSED: Approve minutes of the 22 Mar 2017 Telecon: https://www.w3.org/2017/03/22-shapes-minutes.html


<simonstey> +1

<TallTed> +1

<sandro> +1

<hknublau> +1

RESOLUTION: Approve minutes of the 22 Mar 2017 Telecon: https://www.w3.org/2017/03/22-shapes-minutes.html

<sandro> dallemang: regrets for next week

<TallTed> next call to be Wednesday 2017.04.05, advance regrets from dallemang and dimitris

Dimitris: regrets for next week

public comments

<sandro> discussing thread https://lists.w3.org/Archives/Public/public-rdf-shapes/2017Mar/0068.html

Peter has not replied yet

sandro: there are 2 possible outcomes for this objection: overruled or talk more to Peter
... nothing else we can do about it before the meeting

<sandro> (or some other instructions about what we're supposed to do, but I dunno what)

<TallTed> https://lists.w3.org/Archives/Public/public-rdf-shapes/2017Mar/0078.html

dallemang: hear a lot of confusion from FIBO about SHACL vs ShEx, don't know what to say

sandro: as a user I would want to see what difference is in the tool support, how both look like, what can I do in one vs the other

dallemang: the question I asked was about support, which one is better for FIBO and politics


sandro: Is ShEx about medical informatics and this is the difference we can tell people?

no, not really

Dimitris: should there be a joint statement

TallTed: Yes, a joint statement would be good. Timing of the announcement is questionable

sandro: can we have friendly competition

hknublau: it is hard to compare, SHACL is going through W3C standard and rigorous review, ShEx is in a more flexible situation

Dimitris: I can try to work with the ShEx CG on a joint statement

TallTed: We need a technical co-presentation

there are some differences in capabilities if we compare SHACL Core and ShEx without callouts to Semantic Actions, otherwise both languages are as said "infinite"

dallemang: if SHACL included rules capability from SPIN, then the comparing would have been clear cut

<sandro> (Possible) Outline of Comparison

<sandro> * List of Implementations (with statements about future plans)

<sandro> * List of Other Endorsements (with statements about future plans)

<sandro> * One Example (a Hello World) in both

<sandro> * Examples that show differences

<sandro> (screenshots of tools count like syntax, too)

<sandro> (I'm expecting most people will make up their mind before the detailed exampls)

<sandro> Dimitris: private google document, with comments

<sandro> +1

Dimitris: once there are no more comments, it is endorsed by both groups

TallTed: seems like a good idea, I hesitate about using Google doc

sandro: we could use W3C wiki, although it is not as nice as Google doc for comments, but at least everyone is authenticated

TallTed: yes, W3C wiki is a better way to go

simonstey: for us at Siemens the most important point was that SHACL would be a W3C standard as opposed to a community effort
... PoolParty is in a process of implementing SHACL

<sandro> PoolParty

<simonstey> odrl -> poe

sandro: It is possible for ShEx to become a working group
... The first point in the comparison could be "is it a standard"?

<sandro> sandro: And to get through the W3C process quickly, if they've already done all the work

sandro: Tim will be acting as a director which is very unusual. Normally he delegates
... Tim made an implementation over the weekend, he could ask very technical questions

Test Suite

<simonstey> +q

Is it worthwhile having Andy on standby to answer SPARQL EXISTS questions?

TallTed: yes, this would be great

<simonstey> -q

hknublau: made some namespace changes to the test suite, added test cases, they are in a folder marked as proposed

<simonstey> +q

simonstey: we are too restrictive in declaring conformance

<simonstey> https://github.com/shexSpec/shexTest#validation

simonstey: there are different levels of conformance in ShEx. The simplest one is just pass/fail - logic conformance. There is an experimental part on comparing details of result. Can we go the same way and just go with logic conformance?

I would be in favor of this

TallTed: Good and useful observation. It allows for differentiation between implementations

<TallTed> ack

sandro: are you suggesting a change to specification?
... or is it only about a test suite?

simonstey: test suite currently checks graph isomorphism, I suggest that we only do boolean yes/no comparison for the test suite without changing the spec

<simonstey> http://shexspec.github.io/shexTest/reports/

sandro: may be the simplest thing to do is to say when you reporting results to us distinguish between a) yes, we were able to implement this constraint and correctly check it for pass/fail b) we were able to produce a correct full validation result

simonstey: yes, I am trying to avoid issues that may be raised about graph isomorphism being insufficient

sandro: validation report results may be checked by hand and not only in an automated way

<simonstey> http://shexspec.github.io/shexTest/reports/

<sandro> sandro: Maybe say "automated check LIKELY ALIGNS with the correct answer, but SHOULD be checked by hand"

sandro: if your result is doing graph isomorphism check with this result, your result is highly likely to be correct

<sandro> (to address Peter's corner cases)

TallTed: we may have raised the bar too high in requiring results

<simonstey> http://w3c.github.io/data-shapes/shacl/#validation-report

<dallemang> I got an interrupt that I need to take care of. I need to leave the meeting at this time.

TallTed: can we flip the requirement and say that yes/no is enough to conform?

sandro: it is too late in the process, but it is better do it now than later. It would not help us with the test suite, we still need to test the feature if it exists. But it could help us to get more conforming implementations

<simonstey> +1

sandro: For each test, give us pass/fail and also full results. We will be initially manually checking validation results to see if they converge

<sandro> not "pass/fail" but "shape-match" or "shape-not-match" -- distinction matters for Negative and Positive tests

hknublau: makes sense

<sandro> talking about https://github.com/w3c/data-shapes/tree/gh-pages/data-shapes-test-suite/tests

hknublau: explains folder structure for tests

sandro: makes sense to remove /dash level of the hierarchy

<sandro> Looking at https://github.com/w3c/data-shapes/blob/gh-pages/data-shapes-test-suite/tests/dash/core/complex/personexample.ttl

<simonstey> https://www.w3.org/2013/TurtleTests/?

<sandro> focus: sh:focusNode ex:Alice ;

<sandro> mf:action [

<sandro> sht:dataGraph <> ;

<sandro> sht:shapesGraph <> ;

<sandro> ] ;

sandro: have we said how people should submit test results?

<sandro> https://www.w3.org/2003/08/owl-systems/test-results-out

hknublau: yes, but we need to do it in a more formal, machine readable way

<sandro> http://shexspec.github.io/shexTest/reports/

sandro: how people submit things e.g., e-mail, link to the data and who and how will do the formating of results
... Tim may submit test results
... the simplest thing is a pull request for your EARL data

<sandro> I guess pull-request on your earl output + your validation resuts

<sandro> directory of implemantion reports

<simonstey> sounds reasonable

sandro: you could start doing it for yourself and then people will see it as an example to follow

hknublau: we first need to decide on the tests

sandro: we can have proposed and approved tests, first everything is proposed, then later decide which ones to approve
... could we automatically import some of the ShEx tests?

hknublau: I would be nervous about this
... it is easy to create test cases, if some tests are missing, I would like to hear about theme

simonstey: one test case for each property path

hknublau: created a Wikipedia page, but got a "you are too biased" comment

<simonstey> https://en.wikipedia.org/wiki/SHACL

TallTed: citations to other pages help - like articles that point to the spec

<TallTed> trackbot, end meeting

Summary of Action Items

Summary of Resolutions

  1. Approve minutes of the 22 Mar 2017 Telecon: https://www.w3.org/2017/03/22-shapes-minutes.html
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.152 (CVS log)
$Date: 2017/03/29 13:50:39 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.152  of Date: 2017/02/06 11:04:15  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/@sandro/sandro/
Succeeded: s/@sandro/sandro/
Succeeded: s/PollParty/PoolParty/
Succeeded: s/TallTed,/TallTed:/
Default Present: sandro, simonstey, hknublau, ipolikof, TallTed, Dimitris, dallemang
Present: sandro simonstey hknublau ipolikof TallTed Dimitris dallemang
Regrets: Pano
Found ScribeNick: ipolikof
Inferring Scribes: ipolikof
Found Date: 29 Mar 2017
Guessing minutes URL: http://www.w3.org/2017/03/29-shapes-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]