[
https://issues.apache.org/jira/browse/JENA-1947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17184099#comment-17184099
]
Andy Seaborne commented on JENA-1947:
-------------------------------------
The fix is to remove the line {{sh:and(af-s:AFS_0000001) ;}} . Going by the
{{dct:description}}, it isn't intended. When it says "All Allotrope shapes MUST
reference the debug shape" they mean "All ... except the debug shape itself"!
In a property shape, then the intention might be to walk the path applying a
test every time. That is a possible legitimate usage of recursion.
Jena is a "compiler" style - it reads the shapes, builds an execution tree and
prints or executes that; {{sh:deactivated}] happens at execution.
Other engines are an "interpreter" style. They process the RDF description
while executing - and {{sh:deactivated}} may apply. But I'd not be surprised if
they loop forever if this debug node shape is activated.
Also:
# They don't need to use an "and" of one, {{sh:and(af-s:AFS_0000001)}},
because that's {{sh:node}} if the target of "and" is a node shape.
# A simpler "always triggers" shape is:
{noformat}
:shape2 rdf:type sh:NodeShape ;
sh:targetNode 1 ;
sh:message "ALWAYS" ;
sh:hasValue 2 ;
sh:deactivated true;
.
{noformat}
which is a test that 1 = 2 , which it isn't; or
{noformat}
:shape2 rdf:type sh:NodeShape ;
sh:targetNode [] ;
sh:message "ALWAYS" ;
sh:hasValue [] ;
sh:deactivated true;
.
{noformat}
because the 2 {{[]}} blank nodes are different.
The current SPARQL test does not trigger on an empty graph when data and shapes
are in separate graphs but maybe that is intentional.
> SHACL validator: Stack overflow on recursive shapes even if deactivated
> -----------------------------------------------------------------------
>
> Key: JENA-1947
> URL: https://issues.apache.org/jira/browse/JENA-1947
> Project: Apache Jena
> Issue Type: Bug
> Components: SHACL
> Reporter: Jan Rosecky
> Priority: Major
>
> I'm trying to validate using a set of shapes including:
> [http://purl.allotrope.org/shapes/adm/core/REC/2020/06/common.shapes] .
> af-s:AFS_0000001 has a recursive sh:and(af-s:AFS_0000001) clause, which
> causes a stack overflow exception - even though the shape is sh:deactivated.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
