So "import each other" is not a cycle?
and does the location-mapping.n3 make a difference because I though the
issue was the overridden model.read was not being called which is not
related to location mapping.
Andy
On 22/07/16 21:10, Martynas Jusevičius wrote:
No this is about getOntology() not calling loadImports() in 3.0.1,
which I now need to call myself. And OntModel is retrieved multiple
times.
I see there is Set<String> m_imported in OntModelImpl -- could be
exposed as OntModel.getImportedURIs() for example.
The concurrency thing is another thread. I don't really have more than
those stack-traces. Testing locking OntModel as Dave suggested.
On Fri, Jul 22, 2016 at 10:05 PM, Andy Seaborne <[email protected]> wrote:
Martynas,
This is the concurrency problem? Not JENA-1210 (imports) which happens with
non-cyclic imports?
Could you put a complete, minimal example for this please?
It really does help to have them so that someone can know they are working
on the problem reported. Otherwise, they spend a while just setting up the
situation ... which takes a large chnuk of time ... which means it does not
get done so soon.
Andy
On 22/07/16 18:07, Martynas Jusevičius wrote:
These are 2 ontologies that import each other (gps: imports gp:):
https://github.com/AtomGraph/Processor/blob/master/src/main/resources/org/graphity/processor/gps.ttl
https://github.com/AtomGraph/Processor/blob/master/src/main/resources/org/graphity/processor/gp.ttl
Here is the location mapping (should cover all owl:imports from those
2 ontologies):
https://github.com/AtomGraph/Processor/blob/master/src/main/resources/location-mapping.n3
On Fri, Jul 15, 2016 at 10:01 PM, Andy Seaborne <[email protected]> wrote:
Do you have some test files we can use?
On 15/07/16 20:30, Martynas Jusevičius wrote:
Created an issue: https://issues.apache.org/jira/browse/JENA-1210
On Fri, Jul 15, 2016 at 5:48 PM, Andy Seaborne <[email protected]> wrote:
On 15/07/16 15:15, Martynas Jusevičius wrote:
I am afraid this code requires too much knowledge of OntModel inner
workings, which I don't have. Calling loadImports() explicitly works
as
a
workaround for me.
Wouldn't the fix basically revert to the 2.11.0 code? It did not seem
broken, so why was it "fixed" with RDFDataMgr in the first place?
Because old style RDFReaders now call RDFDataMgr but they can't get the
full
benefit of content negotiation. IIRC.
In 2.11, conneg was weak to non-existent.
A hybrid may be possible.
Andy
On Fri, 15 Jul 2016 at 12:40, Andy Seaborne <[email protected]> wrote:
On 14/07/16 15:17, Martynas Jusevičius wrote:
Should I open a Jira for this as well? Seems like a bug to me.
Yes.
And a fix?
Andy
