Re: Ditching the Constraint Violation Vocabulary

On 3/31/2015 18:47, Dimitris Kontokostas wrote:
>
>     Flexibility always comes at a price. As long as it addresses the
>     use cases, I’d rather take a less complex, less flexible approach.
>
>
> This can be considered a fundamental part of SHACL. Any design choices 
> we make now may affect the ability to improve/revise SHACL in the 
> long-run.

I just bumped into another case supporting Dimitris' view. Each 
constraint may have a field sh:severity, which defaults to sh:Error. A 
SHACL engine can use this information to bypass certain severity levels, 
e.g. if a constraint is marked as sh:Warning then it can be ignored for 
agents only interested in error-level constraints. A proper data 
model/ontology of severities will add flexibility and future-proofness. 
In the current draft the severity levels are classes in an 
rdfs:subClassOf hierarchy.

Holger