- From: Martynas Jusevičius <martynas@graphity.org>
- Date: Wed, 12 Aug 2015 13:37:55 +0200
- To: Phil Archer <phila@w3.org>
- Cc: Miika Alonen <miika.alonen@csc.fi>, kcoyle <kcoyle@kcoyle.net>, Simon Cox <Simon.Cox@csiro.au>, Irene Polikoff <irene@topquadrant.com>, Arnaud Le Hors <lehors@us.ibm.com>, Holger Knublauch <holger@topquadrant.com>, public-data-shapes-wg@w3.org, public-rdf-shapes@w3.org
We actually do the same in our Graphity processor, to match request URIs against resource classes: https://github.com/Graphity/graphity-processor/wiki/How-Graphity-Processor-works But for constraints, I think relying on explicit RDF relationships is a better solution. On Wed, Aug 12, 2015 at 1:17 PM, Phil Archer <phila@w3.org> wrote: > > > On 12/08/2015 12:07, Martynas Jusevičius wrote: >> >> I think syntactical constraints on URIs violate the principle of URI >> opaqueness: http://www.w3.org/TR/webarch/#uri-opacity > > > I don't question that URI opacity is important - but this is a practical > step saying "in this context, I know I need URIs here to match this > pattern." > > And, personally, I've been violating that particular principle - knowingly, > carefully, respectfully - for a while now > http://www.w3.org/TR/powder-grouping/ :-) > > > > > >> >> Martynas >> >> On Wed, Aug 12, 2015 at 11:09 AM, Phil Archer <phila@w3.org> wrote: >>> >>> It might be worth noting that - although failures will happen for any >>> number >>> of reasons - generally speaking if there is a dependency on an external >>> source, those sources are likely to be reference/authority data (things >>> like >>> LOC subject headings, the INSPIRE registry, UK Government time periods >>> etc.). They are going to be stable, predictable and well known to the >>> person >>> writing the constraint. >>> >>> So for me a design that requires the constraint writer to have specific >>> knowledge of what to expect when a specific resource is dereferenced (at >>> any >>> stage in the validation process) is OK. >>> >>> http://id.loc.gov/authorities/subjects/sh85113459, for example, includes >>> the >>> skos:inScheme property and is likely to continue to do so. If that >>> changes - >>> and one has to be prepared for that - then it's going to be a rare event >>> that can be dealt with at that time. The constraint would be that LOC >>> subject heading URIs were used and were genuine and not just a URI that >>> happened to match /^http.*sh\d+$/ >>> >>> Hmmm... >>> >>> Actually, in this case, the test could be: >>> >>> 1. the value of a dcterms:subject property matched >>> /http:\/\/id\.loc\.gov\/authorities\/subjects\/\d+$/ >>> >>> AND >>> >>> 2. an HTTP HEAD request returns a 200 response >>> >>> >>> Phil. >>> >>> >>> >>> On 12/08/2015 09:33, Miika Alonen wrote: >>>> >>>> >>>> Thanks for the reference, its important one! >>>> >>>> I hope resolution would become part of the property constraints where >>>> you >>>> could also describe the severity of the possible resolution failures. It >>>> can >>>> be seen as prevalidation step, but nevertheless it should be considered >>>> as >>>> part of the core spec. >>>> >>>> - Miika >>>> >>>> ----- Original Message ----- >>>> From: "kcoyle" <kcoyle@kcoyle.net> >>>> To: "Miika Alonen" <miika.alonen@csc.fi> >>>> Cc: "Simon Cox" <Simon.Cox@csiro.au>, phila@w3.org, >>>> irene@topquadrant.com, >>>> martynas@graphity.org, lehors@us.ibm.com, holger@topquadrant.com, >>>> public-data-shapes-wg@w3.org, public-rdf-shapes@w3.org >>>> Sent: Tuesday, 11 August, 2015 18:57:05 >>>> Subject: Re: SKOS concept scheme URIs as values for constraints >>>> >>>> On 8/11/15 7:15 AM, Miika Alonen wrote: >>>>> >>>>> >>>>> One general solution would be to support some mechanism for resolving >>>>> resources. I dont know if there has already been discussions about >>>>> dereferencing resources? >>>> >>>> >>>> >>>> >>>> Miika, >>>> >>>> This came up around User Story 40 [1], which was discussed by the group >>>> but which did not result in a specific requirement. The story, provided >>>> by Arthur Ryman (and modified by me), would result in a requirement to >>>> be able to designate specific objects that would need to be resolved >>>> before validation could be applied. (Arthur, if I've got that wrong, pls >>>> correct.) How resolution would be effected was not part of the story. >>>> >>>> Admittedly, validation requiring resolution will be less precise/more >>>> error prone than validation where all data is under ones control. But it >>>> is this less precise world where much academic and cultural heritage >>>> data management takes place. This not only means that we need to resolve >>>> to outside resources, but we need to tolerate some level of failure >>>> without breaking. To me, this is very much in the spirit of RDF. >>>> >>>> kc >>>> [1] >>>> >>>> >>>> http://www.w3.org/2014/data-shapes/wiki/User_Stories#S40_Describing_Inline_Content_versus_References >>>> >>> >>> -- >>> >>> >>> Phil Archer >>> W3C Data Activity Lead >>> http://www.w3.org/2013/data/ >>> >>> http://philarcher.org >>> +44 (0)7887 767755 >>> @philarcher1 >> >> >> > > -- > > > Phil Archer > W3C Data Activity Lead > http://www.w3.org/2013/data/ > > http://philarcher.org > +44 (0)7887 767755 > @philarcher1
Received on Wednesday, 12 August 2015 11:38:27 UTC