Meeting minutes
elianaP: same agenda points as usual, AndyS you added a new agenda point as well
elianaP: no updates on the deliverables from my side
TF updates
HolgerK: I've merged in the unit property, there were no other remarks about this
… I've created a new ticket, issue-725
<AndyS> w3c/
HolgerK: this is a property, sh:intent, which can be used to capture semantics that are otherwise too difficult to express with SHACL
… for many applications I think it's ok to have things like this expressed as text only
… "a parent must be strictly older than each of its children"
… the idea would be to allow people to put those kinds of "rules" in as text first
… and later picked up by someone else to be transformed into an actual shape later one
… or maybe never
… so basically, have a dedicated place where the _intent_ of a shape/rule can be defined in natural language.
bergos: I find this in general useful, but I think there a two aspects to this
… one being: the intent basically representing a natural language description of a constraint
… the other one being: that it represents a constraint that adds to the existing one
… the question is what the intent is supposed to contain
… a description of the (existing) shape itself
… or what it should contain
simonstey: <mentioning sh:message>
AndyS: I was about to mention rdfs:comment, but sh:intent seems to be for the reverse direction
… where the constraint is either non-existent or too week to represent it
HolgerK: yeah I started from that, and can reiterate on that
nicholascar: in geosparql, we had a similar issue where there are geometries typically expressed with coordinates
… but sometimes we would have geometries which we cant express with coordinates, or the coo are not known yet, etc.
… I'm not happy about the term "intent" though
HolgerK: I mean it's just an annotation property
… so please suggest something else in the issue
… in the node expression document there are a lot of todos
… it would be awesome if other editors can have a look at those
… otherwise the document is fairly finished
elianaP: can we get some more examples for the use of sh:intentß
nicholascar: no report from the profiling WG
… I pinged the compact syntax editors, but havent heard back
edmond: SHACL UI, no PR yet
… few ongoing issues
… we meet at our usual times, which is in the calendar
AndyS: not much to report
… got some feedback over the weekend from some N3 folks
… just by looking at the rules you don't necessarily know whether they are satisfiable or not without looking at the data
SHACL protocol
AndyS: we should address all issues where we know that we won't be able to tackle
AndyS: any examples of remote validation?
… the closest I'm aware of would be sending shapes to a processor and it sends you back a report
… where the processor is co-located with the data
… you would have to send the shapes with the request or allow access to shapes to some extent
… [half a protocal is not half a solution ]
Robert: we do have this process, where applications are triggering SHACL validation in the database
… we have a normal REST api
… mentioning RDFox where they added that to SPARQL endpoint with some magic property
… you want to have the validation as close to the data as possible
AndyS: Jena has this sort of REST api as well
… not really advertised though
HolgerK: we offer similar endpoints, via GRAPHQL
… but that's a separate publication I reckon
AndyS: yeah, if someone wants they could provide a Note
AndyS: I'll put a note to the issue
elianaP: that was the last issue I had on the agenda
… any last issues?
<elianaP> rrsagent make minutes