W3C

RDF Data Shapes Working Group Teleconference

28 Jul 2016

Agenda

See also: IRC log

Attendees

Present
hknublau, pano, kcoyle, TallTed, Arnaud, ericP, Dimitris
Regrets
jamsden, andys
Chair
Arnaud
Scribe
dimitris

Contents


<Arnaud> I'm fighting with webex myself...

<scribe> scribenick: dimitris

Admin

<Arnaud> PROPOSED: Approve minutes of the 21 July 2016 Telecon: http://www.w3.org/2016/07/21-shapes-minutes.html

RESOLUTION: Approve minutes of the 21 July 2016 Telecon: http://www.w3.org/2016/07/21-shapes-minutes.html

Arnaud: we meet again next week

Disposal of Raised Issues

<Arnaud> PROPOSED: Open ISSUE-175

<hknublau> +1

Arnaud: Dimitris raised an issue and put together some proposals for renaming the term scope

+1

<kcoyle> +1

RESOLUTION: Open ISSUE-175

scribe: let's open it but should try and close it next week

ISSUE-134

<Arnaud> PROPOSED: Close ISSUE-134 as no longer relevant

Arnaud: Looks like this has become irrelevant with the choices we made

<TallTed> +1

<pano> +1

+1

<kcoyle> +1

kcoyle: have we accepted the proposal for the property path generalization and can anyone point to that?

Dimitris: yes

Kcoyle: I can look up the resolution

<kcoyle> https://www.w3.org/2016/06/30-shapes-minutes.html#resolution04

<kcoyle> we can hear you

<kcoyle> maybe time to retire?

RESOLUTION: Close ISSUE-134 as no longer relevant

<hknublau_> +1

Drafts Publication

Arnaud: When can we publish a new SHACL draft?
... how much time do the editors need?

hknublau: I think issue 133 will have a big impact and then the renaming of scope
... after these are resolved mostly readability and editorial issues will remain

Arnaud: we can decide if we delay 133 and 175

dimitris: I think we should delay a couple of weeks to make the syntax stable

Arnaud: The abstract syntax was improved to tackle the issues we discussed last week but people should check the new document

kcoyle: I would like to hear from people what they think about it

Arnaud: I'd like us to agree on a plan

ericp: there are many buttons that can hide different things in the document

<Zakim> ericP, you wanted to say i'd love to rescue the word "parameters"

Dimitris: I noticed some duplication in the examples, would be nice to align them

<hknublau_> Thanks @ericP to the collapse button for the BNF

ericP: Scoping was a reason for duplication
... would be nice to make the validation work without explicit scopes

Dimitris: There was already an issue about this by Holger

hknublau: It is useful but since we are keep changing at the moment and syncing is not easy

Arnaud: can we somehow automate the syncing

ericP: Scala would be an option

<TallTed> "Nary" should become "N-ary" ... https://en.wikipedia.org/wiki/Arity "Nary" has meanings other than intended.

<kcoyle> thx ted

Arnaud: Holger and Dimitris should signal when you are done so that Eric and Karen can start

Ericp: I will do another pass at naming things
... like parameters / arguments

tallted: if you want to change terminology you should make a complete proposal

Arnaud: I agree with Ted, if you want to rename something make a concrete proposal, at this point we should be very precise

ISSUE-133

<ericP> TallTed, s/nary/n-ary/g done

Arnaud: last week we had a strawpoll and the WG was divided with a slight preference for merging Shapes and NodeConstraints

kcoyle: with Shape becoming a subclass of Constraint it doesn't work for me because we have shapes at different levels
... I need someone to explain to me how this works when constraints contain constraints
... at this point shapes look shapeless

hknublau: we already have this situation anyway, shapes can point to NodeConstraints and NodeConstraints can point to Shapes
... it depends on if we agree if a Shape is actually a big constraints
... it is like an AND construct
... from this point of view merging makes sense

kcoyle: then why do we still have constraint?
... so Shape is a Constraint

hknublau: before we had sh:constraint to link to a NodeConstraint but now we attach directly to the shape

EricP: the problem was with filters I think that was not clear if filters had filters what is happening

<Arnaud> https://www.w3.org/2016/07/21-shapes-minutes.html#item04

<kcoyle> I think that's what I see, too

<Arnaud> STRAWPOLL: a) merge Shape and NodeConstraint, b) don't merge Shape and NodeConstraint

Arnaud: this merge collapses concepts

Dimitris: I also like to keep things separate

<hknublau_> a) +1 b) -0.7

<kcoyle> a) -.9 b) +1

a) -0 b) +1

<TallTed> a +1 b -1

<pano> a:+1 b:0

<ericP> a) -.9

<ericP> a) -.9 b) +1

Arnaud: what do we do with the strawpoll now? Either one would pass because noone has objected firmly
... Eric, how is this related to ShEx?

EricP: this will not change the abstract syntax, this is less about the alignment with ShEx

<Arnaud> PROPOSED: merge Shape and NodeConstraint

<hknublau_> +1

<pano> +1

<TallTed> +1

-0

<kcoyle> -.9

<ericP> -.9

RESOLUTION: Merge Shape and NodeConstraint

Arnaud: with that done we can close issue 133, right?

<Arnaud> PROPOSED: Close ISSUE-133, based on the series of related resolutions

<hknublau_> +1

hknublau: there are a couple of follow-up issues mentioned in my email but can be treated separate

+1

<pano> +1

<kcoyle> +1

RESOLUTION: Close ISSUE-133, based on the series of related resolutions

hknublau: with the merging sh:or is simpler and we can drop sh:datatypeIn and sh:classIn

<Arnaud> https://lists.w3.org/Archives/Public/public-data-shapes-wg/2016Jul/0076.html

ISSUE-135

<trackbot> issue-135 -- Should sh:and/sh:or/sh:not/sh:valueShape support constraints too? -- open

<trackbot> http://www.w3.org/2014/data-shapes/track/issues/135

hknublau: and another issue is if sh:and/or point to constraints which is indirectly resolved

<Arnaud> PROPOSED: Close ISSUE-135, no longer relevant

+1

<pano> +1

<hknublau_> +1

<TallTed> +1

<kcoyle> +1

RESOLUTION: Close ISSUE-135, no longer relevant

ISSUE-141

<trackbot> issue-141 -- How to represent mixed datatype-or-class ranges -- open

<trackbot> http://www.w3.org/2014/data-shapes/track/issues/141

PROPOSED: remove sh:datatypeIn and sh:classIn components as they can be expressed with sh:or

<hknublau_> +1

+1

<ericP> +1

ericP: should we remove sh:in as well

<TallTed> +1

<pano> 0

hknublau: sh:in is useful and I would keep it

<kcoyle> 0

RESOLUTION: Remove sh:datatypeIn and sh:classIn components as they can be expressed with sh:or

<Arnaud> PROPOSED: Close ISSUE-141, mixed ranges can now be uniformly handled

PROPOSED: close issue 141, mixed ranges can be expressed with sh:or

<hknublau_> +1

+1

<TallTed> +1

<ericP> +1

<pano> 0

<kcoyle> +1

RESOLUTION: Close issue 141, mixed ranges can be expressed with sh:or

<hknublau_> Wow, we are on a run. Should we handle the scope naming issue next?

Arnaud: let's talk about nested severities

ISSUE-150

https://lists.w3.org/Archives/Public/public-data-shapes-wg/2016Jul/0079.html

EricP, I had an action item to see how ShEx people would do it and they said they avoid it

<Arnaud> action-37

<trackbot> action-37 -- Eric Prud'hommeaux to Check what happens in the shex extension that has severities -- due 2016-05-26 -- OPEN

<trackbot> http://www.w3.org/2014/data-shapes/track/actions/37

<Arnaud> issue-150

<trackbot> issue-150 -- Treatment of nested severities -- open

<trackbot> http://www.w3.org/2014/data-shapes/track/issues/150

Arnaud: Dimitris sent an email with his preferences

<Arnaud> https://lists.w3.org/Archives/Public/public-data-shapes-wg/2016Jul/0079.html

<kcoyle> lowest = least severe?

b) Use the outer, unless the inner was lower e.g.

outer: Warning, inner: Info -> use Info
... Warning, inner: Violation -> use Warning

kcoyle: I would want the opposite of what Dimitris prefers, the most severe
... if it is inner or outer, the most severe should win

Arnaud: there are different dimensions

kcoyle: but we need to make it very clear

hknublau: I think the current spec is OK for me, if an nested object returns a violation there are many cases where we want to override
... I need some time to look into this in more detail

TallTed: I think the conceptualization of this is key
... I have data about people and require an address
... within the address I can allow a city or state, or more details, if the inner shape is the address
... my address shape must have a street address

<hknublau_> sh:shape may even take a companion parameter to specify the minimum severity.

TallTed: the violation of the inner shape should not mean that the outer shape is broken

<Arnaud> trackbot, end meeting

Summary of Action Items

Summary of Resolutions

  1. Approve minutes of the 21 July 2016 Telecon: http://www.w3.org/2016/07/21-shapes-minutes.html
  2. Open ISSUE-175
  3. Close ISSUE-134 as no longer relevant
  4. Merge Shape and NodeConstraint
  5. Close ISSUE-133, based on the series of related resolutions
  6. Close ISSUE-135, no longer relevant
  7. Remove sh:datatypeIn and sh:classIn components as they can be expressed with sh:or
  8. Close issue 141, mixed ranges can be expressed with sh:or
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.143 (CVS log)
$Date: 2016/08/04 21:17:54 $